Pre-processing

From OLINKS website

Pre-processing and quality control of the data is performed using the Olink NPX Manager software. The built-in quality control (QC) allows us to have control over the technical performance of the assays as well as the samples, making sure that we can deliver reliable, high quality data. The basis of the QC depends on four internal controls that are spiked into all samples and external controls in every Olink Analysis.

The quality control is divided into two parts:

Run evaluation

Standard deviations for Incubation Control 1, Incubation Control 2 and the Detection Control are calculated and should be below the predetermined threshold: 0.2 NPX, for the entire 96-well sample plate.

Sample evaluation

A sample plate median value is calculated for the Incubation Control 2 and the Detection Control, respectively. For each sample, the result for each of these internal controls is allowed to deviate no more than 0.3NPX from the plate median. If any or both of the internal controls exceed the 0.3NPX limit, the sample will fail the QC. If more than 1/6th of the samples fail the QC, the run is deemed unreliable. The reason for the issues will then be evaluated and (if applicable) samples will be rerun.

The NPX Manager software displays the sample evaluation in two different views; the Plate View, showing NPX-values for each sample, and the QC view, showing deviation from the plate median in NPX units. In the example below, all samples but one, “Sample 11” in well B2, passes the QC. For that sample, the Incubation Control 2 and the Detection Control deviate about 0.7 NPX and 1 NPX from the plate median, respectively, and the sample will therefore fail the QC. The NPX values for this sample will be included in the data export file but indicated with red font.

NPX

NPX, Normalized Protein eXpression, is Olink’s arbitrary unit which is in Log2 scale. It is calculated from Ct values and data pre-processing (normalization) is performed to minimize both intra- and inter-assay variation. NPX data allows users to identify changes for individual protein levels across their sample set, and then use this data to establish protein signatures.

NPX is a relative quantification unit logarithmically related to protein concentration. Even if two different proteins have the same NPX values, their absolute concentrations may differ. NPX should be compared for each assay separately between samples within a run. NPX should not be compared between runs without proper inter-plate normalization due to the risk of falsely interpreting shifts in median between runs as a biological difference.

NPX - a difference of 1 NPX equals a doubling of protein concentration

only 1 plate so NO need to cross normalise

Detection threshold

From OLINKS website:

Limit of detection (LOD) is calculated separately for each Olink assay and sample plate. The LOD is based on the background, estimated from negative controls included on every plate, plus three standard deviations. The standard deviation is assay specific and estimated during product validation for every panel.

For studies including more than one plate per panel, the maximum observed LOD for each assay is selected as study LOD. Consequently, all plates included in the study receive the same assay specific LOD. The estimated LOD is a conservative measurement especially in large multiplate studies where there is high probability that observed data is in fact above the true background signal.

Suitable exclusion limits may be in the range of less than 25-50% of samples above LOD.

Characteristics of data below LOD As with all affinity based assays, data from Olink’s platform have a S-curve (sigmoid) relationship with the true protein concentration in a sample. Data below LOD have a higher risk to be in the non-linear phase of the S-curve meaning that 1 NPX difference may not correspond to 2x protein concentration in this region. This may bias estimates including data below LOD and should be considered when interpreting any results that are based on data below LOD

Strategies for handling data below LOD in data analysis Several strategies exist for handling data below LOD that varies in complexity. Olink delivers data below LOD to allow researches to choose the strategy that is best for their study and interpret results with the complexity of data below LOD in mind. Some examples of strategies include:

• Replace data below LOD: It is common to replace data below LOD with a specific value. This will left-censor the data which creates a skewed distribution. Estimates of, for example, mean will be biased and parametric statistical tests may have lower statistical power. Common values to use for replacement is the value for LOD or LOD/(sqrt(2). The latter have been reported in literature to give less biased estimate of means.

• Use actual data below LOD: As data below LOD may be non-linear, estimates of for example mean may be biased. However, especially in large multiplate studies LOD is a conservative measurement. Using actual data may increase the statistical power and give a less skewed distribution compared to replace data below LOD with a value.
• Impute data below LOD: A more complex approach for handling data below LOD is to impute the true value. Several methods for imputation exist and includes maximum-likelihood estimation (MLE) of the distribution below LOD.
• Set data below LOD to missing: Olink do not recommend that data below LOD is excluded from analysis as the most distinct biomarkers may have a low concentration under specific conditions.

Analysis plan:

1. Remove failed samples - rows with NA/blank
2. Remove proteins where expression is < LOD in 50% of samples per group - calculated per group (0,1,2,3). protein need to have low LOD in all groups to be considered for removal
3. Expression values below LOD converted to LOD/sqrt(2). Negative expression values left as is (recommended by OLINK)
4. Set cell values lower than LOD to LOD/(sqrt2) (as recommended by OLINKS)
5. Linear regression - limma:
   • 0 vs 1
   • 0 vs 2
   • 0 vs 3
   • 0 vs 1,2,3 (grouped - case vs control)
   • 0 vs 1 vs 2 vs 3 vs 4
   • longitudinal - 6 patients from mild an 6 patients from severe
6. Visualisation:
   • volcano
   • Boxplot

Read Data set wd

# pc
output_dir<-"D:/Dropbox/Projects/COVID19_blood_biomarker/Analysis"

#laptop
#output_dir<-"C:/Users/hamel/Dropbox/Projects/COVID19_blood_biomarker/Analysis"

# read - pc
protein_data<-read.csv("D:/Dropbox/Projects/COVID19_blood_biomarker/Data/protein_expression_data.txt", sep="\t", head=T, fill=T)
assay_data<-read.csv("D:/Dropbox/Projects/COVID19_blood_biomarker/Data/protein_info.txt", sep="\t", head=T)

# read - laptop
#protein_data<-read.csv("C:/Users/hamel/Dropbox/Projects/COVID19_blood_biomarker/Data/protein_expression_data.txt", sep="\t", head=T, fill=T)
#assay_data<-read.csv("C:/Users/hamel/Dropbox/Projects/COVID19_blood_biomarker/Data/protein_info.txt", sep="\t", head=T)

# dim
dim(protein_data)

## [1]  93 377

Add sample ID

Adding a sample ID to give each sample a unique ID. ID will be “sample” + a number:

1. sample1
2. sample2
3. sample4
4. etc…

# add sample id
rownames(protein_data)<-paste("sample",rownames(protein_data), sep="")

# check
head(protein_data)[1:10]

##         Group Pat.Code Days.since.onset Gender Etnicity   Age   Tau    NfL
## sample1     1      101                8      M     Cauc 71.56 0.880 14.590
## sample2     1      101               27      M     Cauc 71.61 0.990 15.700
## sample3     1      102               12      F     Cauc 71.26 0.918 20.472
## sample4     1      103               22      M     Cauc 69.09 0.188 11.184
## sample5     1      104               23      M     Cauc 66.20 0.808 14.124
## sample6     1      105               17      F     Cauc 60.36 1.810 12.360
##            GFAP OID00379
## sample1 343.830  4.60307
## sample2 484.040  4.89467
## sample3 600.086  4.75002
## sample4 202.817  5.50362
## sample5 279.862  4.72375
## sample6 120.690  4.29556

Groups

Groups:

• 0 = control
• 1 = Mild symptoms
• 2 = Moderate symptoms
• 3 = Severe symptoms (ICU)

The number of samples per group:

table(protein_data$Group)

## 
##  0  1  2  3 
## 33 26  9 25

Number of duplicate samples

Some patients were assayed again at a different time point.

This how many duplicates there are by patient ID:

# count duplicate by pat.code
duplicate_count<-as.data.frame(table(protein_data$Pat.Code))

# count where more than 1
nrow(subset(duplicate_count, Freq > 1))

## [1] 12

# duplicate samples
duplicate_samples<-protein_data[protein_data$Pat.Code %in% subset(duplicate_count, Freq > 1)$Var1,]

# freq by group
table(duplicate_samples[1])

## 
##  1  3 
## 12 12

Calculating time difference between longitudinal data in group 1:

# seperate by group and check difference in Days.since.onset
group1_duplicates_temp<-duplicate_samples[duplicate_samples$Group==1,][c(2,3)]

# data frame to store data
group1_duplicates<-as.data.frame(matrix(nrow=6, ncol=3))

# collapse by duplicate sample ID
temp_counter=1
for (x in seq(1,12,2)) {
    group1_duplicates[temp_counter,c(1,2)]<-group1_duplicates_temp[c(x),]
    group1_duplicates[temp_counter,3]<-group1_duplicates_temp[c(x+1),2]
    temp_counter=temp_counter+1
  }

# remove temp files
rm(temp_counter, group1_duplicates_temp)

# add colnames
colnames(group1_duplicates)<-c("Pat.Code", "1st_sample", "2nd_sample")

# calculate diff
group1_duplicates$time_diff<-as.numeric(group1_duplicates$`2nd_sample`)- as.numeric(group1_duplicates$`1st_sample`)

# table
datatable(group1_duplicates)

# average of diggerence
summary(group1_duplicates$time_diff)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   10.00   14.25   17.00   16.17   19.00   20.00

Calculating time difference between longitudinal data in group 3:

# seperate by group and check difference in Days.since.onset
group3_duplicates_temp<-duplicate_samples[duplicate_samples$Group==3,][c(2,3)]

# data frame to store data
group3_duplicates<-as.data.frame(matrix(nrow=6, ncol=3))

# collapse by duplicate sample ID
temp_counter=1
for (x in seq(1,12,2)) {
  group3_duplicates[temp_counter,c(1,2)]<-group3_duplicates_temp[c(x),]
  group3_duplicates[temp_counter,3]<-group3_duplicates_temp[c(x+1),2]
  temp_counter=temp_counter+1
}

# remove temp files
rm(temp_counter, group3_duplicates_temp)

# add colnames
colnames(group3_duplicates)<-c("Pat.Code", "1st_sample", "2nd_sample")

# calculate diff
group3_duplicates$time_diff<-as.numeric(group3_duplicates$`2nd_sample`)- as.numeric(group3_duplicates$`1st_sample`)

# table
datatable(group3_duplicates)

# average of diggerence
summary(group3_duplicates$time_diff)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   2.000   2.000   3.000   3.167   4.000   5.000

Summary

12 patients have been repeated twice:

• 6 from group 1 (mild)
• 6 from group 3 (severe)

Group 1 average time difference between repeated samples is 16 days

Group 3 avarage time difference between repeated samples is 3 days

Number of duplicate proteins

some proteins repeated on different assay

head(assay_data)

##                             Assay Gene.ID Uniprot.ID OLINK.ID      LOD
## 1 Olink CARDIOVASCULAR II(v.5006)   BMP-6     P22004 OID00379  0.51245
## 2 Olink CARDIOVASCULAR II(v.5006)  ANGPT1     Q15389 OID00380 -0.30886
## 3 Olink CARDIOVASCULAR II(v.5006)     ADM     P35318 OID00381  1.17289
## 4 Olink CARDIOVASCULAR II(v.5006)  CD40-L     P29965 OID00382  0.26247
## 5 Olink CARDIOVASCULAR II(v.5006)  SLAMF7     Q9NQ25 OID00383  1.62544
## 6 Olink CARDIOVASCULAR II(v.5006)     PGF     P49763 OID00384  0.33637

length(unique(assay_data$OLINK.ID))

## [1] 368

length(unique(assay_data$Uniprot.ID))

## [1] 355

length(unique(assay_data$Gene.ID))

## [1] 355

Expression plots

Basic boxplots

# format data
boxplot_data_pre_qc<-stack(as.data.frame(t(protein_data[10:ncol(protein_data)])))

# add group
boxplot_data_pre_qc<-merge(boxplot_data_pre_qc, protein_data[1], by.x="ind", by.y="row.names")

# format
colnames(boxplot_data_pre_qc)<-c("Sample", "NPX", "Group")

# recode group
boxplot_data_pre_qc$Group[boxplot_data_pre_qc$Group=="0"]<-"Control"
boxplot_data_pre_qc$Group[boxplot_data_pre_qc$Group=="1"]<-"Mild"
boxplot_data_pre_qc$Group[boxplot_data_pre_qc$Group=="2"]<-"Moderate"
boxplot_data_pre_qc$Group[boxplot_data_pre_qc$Group=="3"]<-"Severe"

# plot
ggplotly(ggplot(data=boxplot_data_pre_qc, aes(x = Sample, y = NPX, fill=Group)) +
           geom_boxplot() +
           ggtitle("Distribution of protein expression") +
           theme(plot.title = element_text(hjust = 0.5)))

## Warning: Removed 2392 rows containing non-finite values (stat_boxplot).

Density plot

Density plots

# by group
ggplot(boxplot_data_pre_qc, aes(x=NPX, colour=Group)) +
  geom_density() +
  ggtitle("Expression density by Group") +
  theme(plot.title = element_text(hjust = 0.5))

## Warning: Removed 2392 rows containing non-finite values (stat_density).

#by sample
ggplotly(ggplot(boxplot_data_pre_qc, aes(x=NPX, colour=Sample)) +
           geom_density() +
           ggtitle("Expression density by sample") +
           theme(plot.title = element_text(hjust = 0.5), legend.position = "none"))

## Warning: Removed 2392 rows containing non-finite values (stat_density).

Remove samples which are all empty - NA in all rows

There are 4 assays, each with 92 proteins. total = 368

Number of missing prtoein measures by sample (ordered by missingess):

# check samples with mising
sort(rowSums(is.na(protein_data)))

##  sample1  sample3  sample4  sample5  sample6  sample7  sample8  sample9 
##        0        0        0        0        0        0        0        0 
## sample10 sample11 sample12 sample13 sample14 sample15 sample16 sample17 
##        0        0        0        0        0        0        0        0 
## sample18 sample19 sample20 sample21 sample22 sample23 sample24 sample25 
##        0        0        0        0        0        0        0        0 
## sample26 sample27 sample28 sample29 sample30 sample31 sample32 sample33 
##        0        0        0        0        0        0        0        0 
## sample34 sample35 sample36 sample37 sample38 sample39 sample40 sample41 
##        0        0        0        0        0        0        0        0 
## sample42 sample43 sample44 sample45 sample46 sample47 sample48 sample49 
##        0        0        0        0        0        0        0        0 
## sample50 sample52 sample53 sample54 sample55 sample56 sample57 sample58 
##        0        0        0        0        0        0        0        0 
## sample59 sample60 sample61 sample62 sample63 sample64 sample65 sample66 
##        0        0        1        1        1        1        1        1 
## sample67 sample68 sample69 sample70 sample71 sample73 sample74 sample75 
##        1        1        1        1        1        1        1        1 
## sample77 sample78 sample79 sample81 sample82 sample84 sample85 sample86 
##        1        1        1        1        1        1        1        1 
## sample87 sample89 sample90 sample91 sample92 sample93  sample2 sample72 
##        1        1        1        1        1        1      184      369 
## sample76 sample80 sample83 sample88 sample51 
##      369      369      369      369      371

Six samples with missing data and 1 sample with 50% missing (Sample2)

Investigating sample 2:

# find missingness by sample 2
sample_2_missingness<-as.data.frame(t(protein_data[2,10:ncol(protein_data)]))

# merge with assay data
sample_2_missingness<-merge(sample_2_missingness, assay_data[c(1,4)], by.x="row.names", by.y="OLINK.ID")

# change colnames
colnames(sample_2_missingness)<-c("OLINK.ID", "NPX", "Assay")

# change all numeric value to 1
sample_2_missingness[!(is.na(sample_2_missingness$NPX)),2]<-1

# contingency table
table(sample_2_missingness$NPX, sample_2_missingness$Assay)

##    
##     Olink CARDIOVASCULAR II(v.5006) Olink IMMUNE RESPONSE(v.3203)
##   1                              92                             0
##    
##     Olink INFLAMMATION(v.3022) Olink NEUROLOGY(v.8012)
##   1                         92                       0

Summary of sample2:

• Sample2 is a mild symptom patient missing from the immune response and neurology assay (sample kept in)
• It is also a duplicated sample of sample1.
• We can remove this sample for the immune and neurology assay.

Following samples removed (all missing expression values):

1. sample51
2. sample72
3. sample76
4. sample80
5. sample83
6. sample88

# make list of samples to remove
samples_to_remove<-rownames(as.data.frame(tail(sort(rowSums(is.na(protein_data))), 6)))

# The samples eing removed belong to group:
protein_data[rownames(protein_data) %in% samples_to_remove,1]

## [1] 3 0 0 0 0 0

# remove from expression
protein_data_clean<-protein_data[!(rownames(protein_data) %in% samples_to_remove),]

# check
sort(rowSums(is.na(protein_data_clean)))

##  sample1  sample3  sample4  sample5  sample6  sample7  sample8  sample9 
##        0        0        0        0        0        0        0        0 
## sample10 sample11 sample12 sample13 sample14 sample15 sample16 sample17 
##        0        0        0        0        0        0        0        0 
## sample18 sample19 sample20 sample21 sample22 sample23 sample24 sample25 
##        0        0        0        0        0        0        0        0 
## sample26 sample27 sample28 sample29 sample30 sample31 sample32 sample33 
##        0        0        0        0        0        0        0        0 
## sample34 sample35 sample36 sample37 sample38 sample39 sample40 sample41 
##        0        0        0        0        0        0        0        0 
## sample42 sample43 sample44 sample45 sample46 sample47 sample48 sample49 
##        0        0        0        0        0        0        0        0 
## sample50 sample52 sample53 sample54 sample55 sample56 sample57 sample58 
##        0        0        0        0        0        0        0        0 
## sample59 sample60 sample61 sample62 sample63 sample64 sample65 sample66 
##        0        0        1        1        1        1        1        1 
## sample67 sample68 sample69 sample70 sample71 sample73 sample74 sample75 
##        1        1        1        1        1        1        1        1 
## sample77 sample78 sample79 sample81 sample82 sample84 sample85 sample86 
##        1        1        1        1        1        1        1        1 
## sample87 sample89 sample90 sample91 sample92 sample93  sample2 
##        1        1        1        1        1        1      184

List proteins with high LOD rate + NA

Proteins with > 50% LOD in all groups are removed.

Making a list of proteins where more than 50% are below the LOD for each group

#create subset of group 0 (exprs only)
group0_exprs<-(subset(protein_data_clean, protein_data_clean$Group==0))[10:ncol(protein_data_clean)]
dim(group0_exprs)

## [1]  28 368

#create subset of group 1 (exprs only)
group1_exprs<-(subset(protein_data_clean, protein_data_clean$Group==1))[10:ncol(protein_data_clean)]
dim(group1_exprs)

## [1]  26 368

#create subset of group 2 (exprs only)
group2_exprs<-(subset(protein_data_clean, protein_data_clean$Group==2))[10:ncol(protein_data_clean)]
dim(group2_exprs)

## [1]   9 368

#create subset of group 3 (exprs only)
group3_exprs<-(subset(protein_data_clean, protein_data_clean$Group==3))[10:ncol(protein_data_clean)]
dim(group3_exprs)

## [1]  24 368

#check protein order same as assay data
all(colnames(group0_exprs)==assay_data$OLINK.ID)==T

## [1] TRUE

all(colnames(group1_exprs)==assay_data$OLINK.ID)==T

## [1] TRUE

all(colnames(group2_exprs)==assay_data$OLINK.ID)==T

## [1] TRUE

all(colnames(group3_exprs)==assay_data$OLINK.ID)==T

## [1] TRUE

#create count file
missingness<-assay_data[4]

#add groups
missingness$Group0<-0
missingness$Group1<-0
missingness$Group2<-0
missingness$Group3<-0

#count group 0
for (x in 1:nrow(missingness)) {
  missingness[x,2]<-(length(group0_exprs[x][group0_exprs[x]<assay_data$LOD[x]])/nrow(group0_exprs))*100
}

#count group 1
for (x in 1:nrow(missingness)) {
  missingness[x,3]<-(length(group1_exprs[x][group1_exprs[x]<assay_data$LOD[x]])/nrow(group1_exprs))*100
}

#count group 2
for (x in 1:nrow(missingness)) {
  missingness[x,4]<-(length(group2_exprs[x][group2_exprs[x]<assay_data$LOD[x]])/nrow(group2_exprs))*100
}

#count group 3
for (x in 1:nrow(missingness)) {
  missingness[x,5]<-(length(group3_exprs[x][group3_exprs[x]<assay_data$LOD[x]])/nrow(group3_exprs))*100
}

Count the number of proteins where expression values are < LOD per group

This is the count of proteins with > 50% missingness per group.

# move OLINK ID to rowname and remove from file
rownames(missingness)<-missingness$OLINK.ID
missingness$OLINK.ID<-NULL

#count per group
apply(missingness, 2, function(x) {sum(x>50)})

## Group0 Group1 Group2 Group3 
##     17     27     20     18

#count across group where missiness is over 50%
missingness_across_groups<-as.data.frame(apply(missingness, 1, function(x) {sum(x>50)}))
colnames(missingness_across_groups)<-"missingness_count"

# This is the count of 50% across groups.
table(missingness_across_groups)

## missingness_across_groups
##   0   1   2   3   4 
## 341   4   4   6  13

Summary of protein missingness

• 13 proteins have >50% missingness across all groups
• 6 proteins have >50% missingness across 3 groups
• 4 proteins have >50% missingness across 2 groups
• 4 proteins have >50% missingness across 1 groups
• 341 proteins DON’T have >50% missingess in any group

list of proteins to remove. i.e missingness over 50% in all groups

proteins_to_remove<-rownames(subset(missingness_across_groups, missingness_across_groups$missingness_count==4))

Remove proteins with high LOD + NA across all groups

Remove the 13 proteins from expression

# proteins to remove
length(proteins_to_remove)

## [1] 13

# number of proteins
length(colnames(protein_data_clean[10:ncol(protein_data_clean)]))

## [1] 368

# remove proteins
protein_data_clean<-protein_data_clean[!(colnames(protein_data_clean) %in% proteins_to_remove)]

# number of proteins
length(colnames(protein_data_clean[10:ncol(protein_data_clean)]))

## [1] 355

Remove proteins from assay data

# number of proteins
length(rownames(assay_data))

## [1] 368

# remove proteins
assay_data_clean<-assay_data[!(assay_data$OLINK.ID %in% proteins_to_remove),]

# number of proteins
length(rownames(assay_data_clean))

## [1] 355

check expression data and assay data has same proteins

all(assay_data_clean$OLINK.ID==colnames(protein_data_clean[10:ncol(protein_data_clean)]))==T

## [1] TRUE

Replace low expression

How many expression values < LOD in data:

# extract expression table only
exprs_only<-protein_data_clean[10:ncol(protein_data_clean)]

# check protein order same
all(colnames(exprs_only)==assay_data_clean$OLINK.ID)

## [1] TRUE

# count total number of cells with value less than its LOD
total_low_LOD_count<-assay_data_clean[4]
total_low_LOD_count$missingness<-0

# individual count
for (x in 1:nrow(total_low_LOD_count)) {
  total_low_LOD_count[x,2]<-length(na.omit(exprs_only[x][exprs_only[x] < assay_data_clean$LOD[x]]))
}

# total number of expression values below LOD
sum(total_low_LOD_count$missingness)

## [1] 834

# total number of proteins with an expression value less than LOD
nrow(subset(total_low_LOD_count, total_low_LOD_count$missingness!=0))

## [1] 44

# number of proteins with an expression value less than LOD - what panels are they in
table(assay_data_clean[assay_data_clean$OLINK.ID %in% subset(total_low_LOD_count, total_low_LOD_count$missingness!=0)$OLINK.ID,]$Assay)

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                               2                              24 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                              13                               5

Change these low LOD values to LOD/sqrt(2)

# for each column, replace values lower than LOD to LOD/sqrt(2) 
for (x in 1:ncol(exprs_only)) {
  exprs_only[x]<-replace(exprs_only[x], 
                   exprs_only[x] < assay_data_clean$LOD[x], 
                   assay_data_clean$LOD[x]/(sqrt(2)))
}

Check again how many value are less than LOD. This time have to check if values is less than the new assigned value LOD/sqrt(2)

# count less than LOD/sqrt(2)
total_low_LOD_count$missingness2<-0

# individual count
for (x in 1:nrow(total_low_LOD_count)) {
  total_low_LOD_count[x,3]<-length(na.omit(exprs_only[x][exprs_only[x] < assay_data_clean$LOD[x]/(sqrt(2))]))
  }

# count
sum(total_low_LOD_count$missingness2)

## [1] 20

Replace expression data in clean data

#replace exression data
protein_data_clean[10:ncol(protein_data_clean)]<-exprs_only

#check coloumns still in order
all(colnames(protein_data_clean)[10:ncol(protein_data_clean)]==assay_data_clean$OLINK.ID)==T

## [1] TRUE

PCA

Performing PCA prior to data processing. PCA does not handle missing values, so proteins with missingnes are removed.

# pca - remove NAs - PCA does not like them
pca_data_na_removed<-na.omit(as.data.frame(t(protein_data_clean[10:ncol(protein_data_clean)])))

# numbr of proteins remaining 
nrow(pca_data_na_removed)

## [1] 173

# run PCA
pca_data<-prcomp(pca_data_na_removed)

# summary
summary(pca_data)

## Importance of components:
##                           PC1    PC2     PC3     PC4     PC5     PC6     PC7
## Standard deviation     27.672 5.9117 3.69170 1.92258 1.63432 1.46674 1.33791
## Proportion of Variance  0.905 0.0413 0.01611 0.00437 0.00316 0.00254 0.00212
## Cumulative Proportion   0.905 0.9463 0.96239 0.96676 0.96992 0.97246 0.97458
##                            PC8     PC9   PC10    PC11    PC12    PC13   PC14
## Standard deviation     1.27822 1.18915 1.0893 1.03854 0.98079 0.95904 0.9211
## Proportion of Variance 0.00193 0.00167 0.0014 0.00127 0.00114 0.00109 0.0010
## Cumulative Proportion  0.97651 0.97818 0.9796 0.98086 0.98199 0.98308 0.9841
##                           PC15    PC16   PC17    PC18   PC19    PC20    PC21
## Standard deviation     0.88753 0.88417 0.8244 0.78928 0.7684 0.70312 0.69719
## Proportion of Variance 0.00093 0.00092 0.0008 0.00074 0.0007 0.00058 0.00057
## Cumulative Proportion  0.98501 0.98594 0.9867 0.98748 0.9882 0.98876 0.98933
##                           PC22    PC23    PC24    PC25    PC26    PC27    PC28
## Standard deviation     0.69037 0.66786 0.65778 0.62561 0.60703 0.59309 0.59044
## Proportion of Variance 0.00056 0.00053 0.00051 0.00046 0.00044 0.00042 0.00041
## Cumulative Proportion  0.98990 0.99042 0.99093 0.99140 0.99183 0.99225 0.99266
##                           PC29    PC30    PC31    PC32    PC33    PC34    PC35
## Standard deviation     0.56691 0.56314 0.54632 0.53285 0.52758 0.51224 0.49714
## Proportion of Variance 0.00038 0.00037 0.00035 0.00034 0.00033 0.00031 0.00029
## Cumulative Proportion  0.99304 0.99342 0.99377 0.99410 0.99443 0.99474 0.99503
##                           PC36    PC37    PC38    PC39    PC40    PC41    PC42
## Standard deviation     0.48182 0.46702 0.45864 0.43671 0.43017 0.42053 0.41861
## Proportion of Variance 0.00027 0.00026 0.00025 0.00023 0.00022 0.00021 0.00021
## Cumulative Proportion  0.99531 0.99557 0.99582 0.99604 0.99626 0.99647 0.99668
##                           PC43    PC44    PC45    PC46    PC47    PC48    PC49
## Standard deviation     0.40589 0.39303 0.38133 0.36610 0.36004 0.35717 0.35053
## Proportion of Variance 0.00019 0.00018 0.00017 0.00016 0.00015 0.00015 0.00015
## Cumulative Proportion  0.99687 0.99705 0.99722 0.99738 0.99754 0.99769 0.99783
##                           PC50    PC51    PC52    PC53    PC54   PC55   PC56
## Standard deviation     0.34295 0.33212 0.31615 0.31093 0.30086 0.2962 0.2927
## Proportion of Variance 0.00014 0.00013 0.00012 0.00011 0.00011 0.0001 0.0001
## Cumulative Proportion  0.99797 0.99810 0.99822 0.99833 0.99844 0.9985 0.9987
##                           PC57    PC58    PC59    PC60    PC61    PC62    PC63
## Standard deviation     0.27543 0.27354 0.27234 0.25679 0.25171 0.24003 0.23846
## Proportion of Variance 0.00009 0.00009 0.00009 0.00008 0.00007 0.00007 0.00007
## Cumulative Proportion  0.99874 0.99882 0.99891 0.99899 0.99906 0.99913 0.99920
##                           PC64    PC65    PC66    PC67    PC68    PC69    PC70
## Standard deviation     0.23449 0.23147 0.22204 0.21257 0.20800 0.20387 0.19496
## Proportion of Variance 0.00006 0.00006 0.00006 0.00005 0.00005 0.00005 0.00004
## Cumulative Proportion  0.99926 0.99933 0.99939 0.99944 0.99949 0.99954 0.99959
##                           PC71    PC72    PC73    PC74    PC75    PC76    PC77
## Standard deviation     0.19228 0.18889 0.17939 0.16721 0.16099 0.15834 0.15376
## Proportion of Variance 0.00004 0.00004 0.00004 0.00003 0.00003 0.00003 0.00003
## Cumulative Proportion  0.99963 0.99967 0.99971 0.99974 0.99977 0.99980 0.99983
##                           PC78    PC79    PC80    PC81    PC82    PC83    PC84
## Standard deviation     0.14866 0.13920 0.13673 0.12891 0.12542 0.11510 0.10475
## Proportion of Variance 0.00003 0.00002 0.00002 0.00002 0.00002 0.00002 0.00001
## Cumulative Proportion  0.99986 0.99988 0.99990 0.99992 0.99994 0.99996 0.99997
##                           PC85    PC86    PC87
## Standard deviation     0.09805 0.09393 0.09216
## Proportion of Variance 0.00001 0.00001 0.00001
## Cumulative Proportion  0.99998 0.99999 1.00000

# extract PC1 and PC2
PC1_plot<-pca_data$rotation[,1:2]

# add phenotype information to PC information
PC1_plot<-cbind(PC1_plot, protein_data_clean[c(1,3,4,5)])
PC1_plot$Group<-as.character(PC1_plot$Group)

# ggplot plot 
ggplotly(ggplot(data=PC1_plot, aes(x=PC1, y=PC2, col=Group)) +
  # scatterplot
  geom_point() +
  # add title to plot
  ggtitle("PC1 vs PC2 by Group") +
  # centre title 
  theme(plot.title = element_text(hjust = 0.5)))

plot by Ethnicity

# ggplot plot 
ggplotly(ggplot(data=PC1_plot, aes(x=PC1, y=PC2, col=Etnicity)) +
  # scatterplot
  geom_point() +
  # add title to plot
  ggtitle("PC1 vs PC2 by Ethnicity") +
  # centre title 
  theme(plot.title = element_text(hjust = 0.5)))

Ethinicity for all controls are uknown

plot by Gender

# ggplot plot 
ggplotly(ggplot(data=PC1_plot, aes(x=PC1, y=PC2, col=Days.since.onset)) +
  # scatterplot
  geom_point() +
  # add title to plot
  ggtitle("PC1 vs PC2 by Days.since.onset") +
  # centre title 
  theme(plot.title = element_text(hjust = 0.5)))

Expression boxplot

Data expression plot aftr QC

Basic boxplots

# format data
boxplot_data_post_qc<-stack(as.data.frame(t(protein_data_clean[10:ncol(protein_data_clean)])))

# add group
boxplot_data_post_qc<-merge(boxplot_data_post_qc, protein_data_clean[1], by.x="ind", by.y="row.names")

# format
colnames(boxplot_data_post_qc)<-c("Sample", "NPX", "Group")

# recode group
boxplot_data_post_qc$Group[boxplot_data_post_qc$Group=="0"]<-"Control"
boxplot_data_post_qc$Group[boxplot_data_post_qc$Group=="1"]<-"Mild"
boxplot_data_post_qc$Group[boxplot_data_post_qc$Group=="2"]<-"Moderate"
boxplot_data_post_qc$Group[boxplot_data_post_qc$Group=="3"]<-"Severe"

# plot
ggplotly(ggplot(data=boxplot_data_post_qc, aes(x = Sample, y = NPX, fill=Group)) +
           geom_boxplot() +
           ggtitle("Distribution of protein expression") +
           theme(plot.title = element_text(hjust = 0.5)))

## Warning: Removed 182 rows containing non-finite values (stat_boxplot).

Density plot

Density plots

# by group
ggplot(boxplot_data_post_qc, aes(x=NPX, colour=Group)) +
  geom_density() +
  ggtitle("Expression density by Group") +
  theme(plot.title = element_text(hjust = 0.5))

## Warning: Removed 182 rows containing non-finite values (stat_density).

#by sample
ggplotly(ggplot(boxplot_data_post_qc, aes(x=NPX, colour=Sample)) +
           geom_density() +
           ggtitle("Expression density by sample") +
           theme(plot.title = element_text(hjust = 0.5), legend.position = "none"))

## Warning: Removed 182 rows containing non-finite values (stat_density).

Groups

Groups:

• 0 = control
• 1 = Mild symptoms
• 2 = Moderate symptoms
• 3 = Severe symptoms (ICU)

The number of samples per group:

table(protein_data_clean$Group)

## 
##  0  1  2  3 
## 28 26  9 24

Age

• summarise all into a table

summary of age

# overall age
summary(protein_data_clean$Age)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   21.52   47.42   59.67   59.18   71.41   87.29

# 95 CI by group 

CI(subset(protein_data_clean, protein_data_clean$Group==0)$Age, ci=0.95)

##    upper     mean    lower 
## 70.27734 63.10643 55.93552

CI(subset(protein_data_clean, protein_data_clean$Group==1)$Age, ci=0.95)

##    upper     mean    lower 
## 56.79125 51.34500 45.89875

CI(subset(protein_data_clean, protein_data_clean$Group==2)$Age, ci=0.95)

##    upper     mean    lower 
## 74.63664 64.82222 55.00780

CI(subset(protein_data_clean, protein_data_clean$Group==3)$Age, ci=0.95)

##    upper     mean    lower 
## 65.79818 60.98125 56.16432

# CI for all cases
CI(subset(protein_data_clean, !(protein_data_clean$Group==0))$Age, ci=0.95)

##    upper     mean    lower 
## 60.84354 57.32068 53.79782

ggplot(data=protein_data_clean, aes(x=as.character(Group), y=Age, fill=Group)) +
  geom_boxplot() +
  ggtitle("Distribution of age by group") +
  theme(plot.title = element_text(hjust = 0.5))

T-test of Group 0 vs 1 (control vs mild)

# group 0 vs 1
t.test(protein_data_clean[protein_data_clean$Group==0,6], protein_data_clean[protein_data_clean$Group==1,6])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 0, 6] and protein_data_clean[protein_data_clean$Group == 1, 6]
## t = 2.6837, df = 49.31, p-value = 0.009886
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##   2.955688 20.567170
## sample estimates:
## mean of x mean of y 
##  63.10643  51.34500

T-test of Group 0 vs 2 (control vs moderate)

# group 0 vs 2
t.test(protein_data_clean[protein_data_clean$Group==0,6], protein_data_clean[protein_data_clean$Group==2,6])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 0, 6] and protein_data_clean[protein_data_clean$Group == 2, 6]
## t = -0.31156, df = 19.764, p-value = 0.7586
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -13.212180   9.780593
## sample estimates:
## mean of x mean of y 
##  63.10643  64.82222

T-test of Group 0 vs 3 (control vs severe)

# group 0 vs 3
t.test(protein_data_clean[protein_data_clean$Group==0,6], protein_data_clean[protein_data_clean$Group==3,6])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 0, 6] and protein_data_clean[protein_data_clean$Group == 3, 6]
## t = 0.50605, df = 45.716, p-value = 0.6153
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -6.329499 10.579856
## sample estimates:
## mean of x mean of y 
##  63.10643  60.98125

T-test of Group 1 vs 2 (mild vs moderate)

# group 1 vs 2
t.test(protein_data_clean[protein_data_clean$Group==1,6], protein_data_clean[protein_data_clean$Group==2,6])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 1, 6] and protein_data_clean[protein_data_clean$Group == 2, 6]
## t = -2.6897, df = 14.67, p-value = 0.01705
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -24.178172  -2.776273
## sample estimates:
## mean of x mean of y 
##  51.34500  64.82222

T-test of Group 1 vs 3 (mild vs severe)

# group 1 vs 3
t.test(protein_data_clean[protein_data_clean$Group==1,6], protein_data_clean[protein_data_clean$Group==3,6])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 1, 6] and protein_data_clean[protein_data_clean$Group == 3, 6]
## t = -2.7349, df = 47.656, p-value = 0.008736
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -16.722004  -2.550496
## sample estimates:
## mean of x mean of y 
##  51.34500  60.98125

T-test of Group 2 vs 3 (moderate vs severe)

# group 2 vs 3
t.test(protein_data_clean[protein_data_clean$Group==2,6], protein_data_clean[protein_data_clean$Group==3,6])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 2, 6] and protein_data_clean[protein_data_clean$Group == 3, 6]
## t = 0.79173, df = 13.098, p-value = 0.4426
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -6.631821 14.313766
## sample estimates:
## mean of x mean of y 
##  64.82222  60.98125

T-test of Group 0 vs 1,2,3 (control vs case)

# group 0 vs 1,2,3
t.test(protein_data_clean[protein_data_clean$Group==0,6], protein_data_clean[protein_data_clean$Group %in% c(1,2,3),6])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 0, 6] and protein_data_clean[protein_data_clean$Group %in% c(1, 2, 3), protein_data_clean[protein_data_clean$Group == 0, 6] and     6]
## t = 1.4786, df = 41.196, p-value = 0.1469
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -2.115556 13.687058
## sample estimates:
## mean of x mean of y 
##  63.10643  57.32068

Summary of Age

• Significant difference between group 0 and 1 only
• no significant differnce when comparing controls against all cases (grouped)

Gender

• summarise all into a table (Fisher is better for smaller numbers)

summary of Gender

# overall Gender
table(protein_data_clean$Gender)

## 
##  F  M 
## 31 56

# gender by disease group
table(protein_data_clean$Gender, protein_data_clean$Group)

##    
##      0  1  2  3
##   F 13 13  3  2
##   M 15 13  6 22

Fisher test of Group 0 vs 1 (control vs mild)

# extract
group_0_vs_1_GG<-protein_data_clean[protein_data_clean$Group==0 | protein_data_clean$Group==1,]
# test
fisher.test(table(group_0_vs_1_GG$Group, group_0_vs_1_GG$Gender))

## 
##  Fisher's Exact Test for Count Data
## 
## data:  table(group_0_vs_1_GG$Group, group_0_vs_1_GG$Gender)
## p-value = 1
## alternative hypothesis: true odds ratio is not equal to 1
## 95 percent confidence interval:
##  0.2609774 2.8733805
## sample estimates:
## odds ratio 
##  0.8689712

Fisher test of Group 0 vs 2 (control vs moderate)

# extract
group_0_vs_2_GG<-protein_data_clean[protein_data_clean$Group==0 | protein_data_clean$Group==2,]
# test
fisher.test(table(group_0_vs_2_GG$Group, group_0_vs_2_GG$Gender))

## 
##  Fisher's Exact Test for Count Data
## 
## data:  table(group_0_vs_2_GG$Group, group_0_vs_2_GG$Gender)
## p-value = 0.7023
## alternative hypothesis: true odds ratio is not equal to 1
## 95 percent confidence interval:
##   0.2908013 12.7086492
## sample estimates:
## odds ratio 
##   1.708209

Fisher test of Group 0 vs 3 (control vs severe)

# extract
group_0_vs_3_GG<-protein_data_clean[protein_data_clean$Group==0 | protein_data_clean$Group==3,]
# test
fisher.test(table(group_0_vs_3_GG$Group, group_0_vs_3_GG$Gender))

## 
##  Fisher's Exact Test for Count Data
## 
## data:  table(group_0_vs_3_GG$Group, group_0_vs_3_GG$Gender)
## p-value = 0.004729
## alternative hypothesis: true odds ratio is not equal to 1
## 95 percent confidence interval:
##   1.696152 95.007720
## sample estimates:
## odds ratio 
##   9.132736

Fisher test of Group 1 vs 2 (control vs severe)

# extract
group_1_vs_2_GG<-protein_data_clean[protein_data_clean$Group==1 | protein_data_clean$Group==2,]
# test
fisher.test(table(group_1_vs_2_GG$Group, group_1_vs_2_GG$Gender))

## 
##  Fisher's Exact Test for Count Data
## 
## data:  table(group_1_vs_2_GG$Group, group_1_vs_2_GG$Gender)
## p-value = 0.4605
## alternative hypothesis: true odds ratio is not equal to 1
## 95 percent confidence interval:
##   0.3294235 14.7840643
## sample estimates:
## odds ratio 
##   1.961239

Fisher test of Group 1 vs 3 (control vs severe)

# extract
group_1_vs_3_GG<-protein_data_clean[protein_data_clean$Group==1 | protein_data_clean$Group==3,]
# test
fisher.test(table(group_1_vs_3_GG$Group, group_1_vs_3_GG$Gender))

## 
##  Fisher's Exact Test for Count Data
## 
## data:  table(group_1_vs_3_GG$Group, group_1_vs_3_GG$Gender)
## p-value = 0.001765
## alternative hypothesis: true odds ratio is not equal to 1
## 95 percent confidence interval:
##    1.916874 110.135285
## sample estimates:
## odds ratio 
##   10.46825

Fisher test of Group 2 vs 3 (control vs severe)

# extract
group_2_vs_3_GG<-protein_data_clean[protein_data_clean$Group==2 | protein_data_clean$Group==3,]
# test
fisher.test(table(group_2_vs_3_GG$Group, group_2_vs_3_GG$Gender))

## 
##  Fisher's Exact Test for Count Data
## 
## data:  table(group_2_vs_3_GG$Group, group_2_vs_3_GG$Gender)
## p-value = 0.111
## alternative hypothesis: true odds ratio is not equal to 1
## 95 percent confidence interval:
##   0.4776759 75.1620727
## sample estimates:
## odds ratio 
##    5.15309

Fisher test of Group 0 vs 1,2,3 (control vs case)

# extract
case_vs_control_GG<-protein_data_clean
case_vs_control_GG$Group<-ifelse(case_vs_control_GG$Group==0, "control", "case")

# test
fisher.test(table(case_vs_control_GG$Group, case_vs_control_GG$Gender))

## 
##  Fisher's Exact Test for Count Data
## 
## data:  table(case_vs_control_GG$Group, case_vs_control_GG$Gender)
## p-value = 0.1596
## alternative hypothesis: true odds ratio is not equal to 1
## 95 percent confidence interval:
##  0.1820679 1.4254448
## sample estimates:
## odds ratio 
##  0.5107209

Summary of Gender

• Signiicantly More Males than females in 0 vs 3, 1 vs 3
• no significant differnce when comparing controls against all cases (grouped)

Ethnicity

Ethnicity by group

table(protein_data_clean$Etnicity, protein_data_clean$Group)

##                
##                  0  1  2  3
##                 28  0  0  0
##   African black  0  0  0  1
##   Cauc           0 20  8 23
##   Persian        0  6  1  0

Number of duplicate samples

Check if we removed any duplicates

# count duplicate by pat.code
duplicate_count_after_qc<-as.data.frame(table(protein_data_clean$Pat.Code))

# count where more than 1
nrow(subset(duplicate_count_after_qc, Freq > 1))

## [1] 11

# duplicate samples
duplicate_samples_after_qc<-protein_data_clean[protein_data_clean$Pat.Code %in% subset(duplicate_count_after_qc, Freq > 1)$Var1,]

# freq by group
table(duplicate_samples_after_qc[1])

## 
##  1  3 
## 12 10

# extract sample info
datatable(duplicate_samples_after_qc[1:4], options = list(pageLength = 24))

Summary of duplicates after QC

• one set of duplicates lost from group3.
• Group1: 12 samples (6 patients)
• Group3: 10 samples (5 patients)

Number of duplicate proteins

some proteins repeated on different assay

head(assay_data)

##                             Assay Gene.ID Uniprot.ID OLINK.ID      LOD
## 1 Olink CARDIOVASCULAR II(v.5006)   BMP-6     P22004 OID00379  0.51245
## 2 Olink CARDIOVASCULAR II(v.5006)  ANGPT1     Q15389 OID00380 -0.30886
## 3 Olink CARDIOVASCULAR II(v.5006)     ADM     P35318 OID00381  1.17289
## 4 Olink CARDIOVASCULAR II(v.5006)  CD40-L     P29965 OID00382  0.26247
## 5 Olink CARDIOVASCULAR II(v.5006)  SLAMF7     Q9NQ25 OID00383  1.62544
## 6 Olink CARDIOVASCULAR II(v.5006)     PGF     P49763 OID00384  0.33637

length(unique(assay_data_clean$OLINK.ID))

## [1] 355

length(unique(assay_data_clean$Uniprot.ID))

## [1] 344

length(unique(assay_data_clean$Gene.ID))

## [1] 344

Time difference for longitudinal data

Calculating time difference between longitudinal data in group 1

# seperate by group and check difference in Days.since.onset
group1_duplicates_after_qc_temp2<-duplicate_samples_after_qc[duplicate_samples_after_qc$Group==1,][c(2,3)]

# data frame to store data
group1_duplicates_after_qc<-as.data.frame(matrix(nrow=6, ncol=3))

# collapse by duplicate sample ID
temp_counter=1
for (x in seq(1,12,2)) {
  group1_duplicates_after_qc[temp_counter,c(1,2)]<-group1_duplicates_after_qc_temp2[c(x),]
  group1_duplicates_after_qc[temp_counter,3]<-group1_duplicates_after_qc_temp2[c(x+1),2]
  temp_counter=temp_counter+1
}

# remove temp files
rm(temp_counter, group1_duplicates_after_qc_temp2)

# add colnames
colnames(group1_duplicates_after_qc)<-c("Pat.Code", "1st_sample", "2nd_sample")

# calculate diff
group1_duplicates_after_qc$time_diff<-as.numeric(group1_duplicates_after_qc$`2nd_sample`)- as.numeric(group1_duplicates_after_qc$`1st_sample`)

# table
datatable(group1_duplicates_after_qc)

This is the time difference in repeat sampling in group 1

# average of difference
summary(group1_duplicates_after_qc$time_diff)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##   10.00   14.25   17.00   16.17   19.00   20.00

Calculating time difference between longitudinal data in group 3

# seperate by group and check difference in Days.since.onset
group3_duplicates_after_qc_temp<-duplicate_samples_after_qc[duplicate_samples_after_qc$Group==3,][c(2,3)]

# data frame to store data
group3_duplicates_after_qc<-as.data.frame(matrix(nrow=5, ncol=3))

# collapse by duplicate sample ID
temp_counter=1
for (x in seq(1,10,2)) {
  group3_duplicates_after_qc[temp_counter,c(1,2)]<-group3_duplicates_after_qc_temp[c(x),]
  group3_duplicates_after_qc[temp_counter,3]<-group3_duplicates_after_qc_temp[c(x+1),2]
  temp_counter=temp_counter+1
}

# remove temp files
rm(temp_counter, group3_duplicates_after_qc_temp)

# add colnames
colnames(group3_duplicates_after_qc)<-c("Pat.Code", "1st_sample", "2nd_sample")

# calculate diff
group3_duplicates_after_qc$time_diff<-as.numeric(group3_duplicates_after_qc$`2nd_sample`)- as.numeric(group3_duplicates_after_qc$`1st_sample`)

# table
datatable(group3_duplicates_after_qc)

This is the time difference in repeat sampling in group 3

# average of difference
summary(group3_duplicates_after_qc$time_diff)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max. 
##     2.0     2.0     2.0     2.8     4.0     4.0

Summary

11 patients have been repeated twice:

• 6 from group 1 (mild)
• 5 from group 3 (severe)
• Group 1 average time difference between repeated samples is 16 days
• Group 3 avarage time difference between repeated samples is 3 days

summary of demographics in logitudinal patients

summary of demographics of longitudinal patients

group1_duplicates_demographics<-subset(protein_data_clean, protein_data_clean$Pat.Code %in% group1_duplicates_after_qc$Pat.Code)[c(2, 4,5,6)]
group1_duplicates_demographics<-group1_duplicates_demographics[!duplicated(group1_duplicates_demographics$Pat.Code),]

#age
CI(group1_duplicates_demographics$Age, ci=0.95)

##    upper     mean    lower 
## 66.16550 51.48167 36.79783

# gender/ethnicity
group1_duplicates_demographics

##          Pat.Code Gender Etnicity   Age
## sample1       101      M     Cauc 71.56
## sample6       105      F     Cauc 60.36
## sample13      111      F     Cauc 55.44
## sample15      112      M     Cauc 48.02
## sample19      115      M  Persian 40.97
## sample24      119      F  Persian 32.54

# sampling time
CI(group1_duplicates_after_qc$`1st_sample`, ci=0.95)

##     upper      mean     lower 
## 13.587328  8.500000  3.412672

CI(group1_duplicates_after_qc$`2nd_sample`, ci=0.95)

##    upper     mean    lower 
## 27.12034 24.66667 22.21299

CI(group1_duplicates_after_qc$time_diff, ci=0.95)

##    upper     mean    lower 
## 20.22659 16.16667 12.10674

group3_duplicates_demographics<-subset(protein_data_clean, protein_data_clean$Pat.Code %in% group3_duplicates_after_qc$Pat.Code)[c(2, 4,5,6)]
group3_duplicates_demographics<-group3_duplicates_demographics[!duplicated(group3_duplicates_demographics$Pat.Code),]

# age
CI(group3_duplicates_demographics$Age, ci=0.95)

##    upper     mean    lower 
## 73.40117 59.59000 45.77883

# gender/ethnicity

# sampling time

CI(group3_duplicates_after_qc$`1st_sample`, ci=0.95)

##     upper      mean     lower 
## 11.258525  8.400000  5.541475

CI(group3_duplicates_after_qc$`2nd_sample`, ci=0.95)

##     upper      mean     lower 
## 15.067141 11.200000  7.332859

CI(group3_duplicates_after_qc$time_diff, ci=0.95)

##    upper     mean    lower 
## 4.160175 2.800000 1.439825

Days since onset with longitudinal

calculate 95% CI

CI(subset(protein_data_clean, protein_data_clean$Group==0)$Days.since.onset, ci=0.95)

## upper  mean lower 
##    NA    NA    NA

CI(subset(protein_data_clean, protein_data_clean$Group==1)$Days.since.onset, ci=0.95)

##    upper     mean    lower 
## 24.49607 20.88462 17.27316

CI(subset(protein_data_clean, protein_data_clean$Group==2)$Days.since.onset, ci=0.95)

##    upper     mean    lower 
## 13.71879 12.00000 10.28121

CI(subset(protein_data_clean, protein_data_clean$Group==3)$Days.since.onset, ci=0.95)

##     upper      mean     lower 
## 13.008426 11.041667  9.074907

# case s control
CI(subset(protein_data_clean, !(protein_data_clean$Group==0))$Days.since.onset, ci=0.95)

##    upper     mean    lower 
## 17.65654 15.52542 13.39431

summary

Exploration of days since onset. Control group have been set 0

# summary
summary(protein_data_clean$Days.since.onset)

##    Min. 1st Qu.  Median    Mean 3rd Qu.    Max.    NA's 
##    2.00    8.50   13.00   15.53   22.00   35.00      28

# set controls to 0
protein_data_clean[is.na(protein_data_clean$Days.since.onset),3]<-0

#plot
ggplot(data=protein_data_clean, aes(x=as.character(Group), y=Days.since.onset, fill=Group)) +
  geom_boxplot() +
  ggtitle("Distribution of onset_days by group") +
  theme(plot.title = element_text(hjust = 0.5))

T-test of Group 1 vs 2 (control vs mild)

# group 1 vs 2
t.test(protein_data_clean[protein_data_clean$Group==1,3], protein_data_clean[protein_data_clean$Group==2,3])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 1, 3] and protein_data_clean[protein_data_clean$Group == 2, 3]
## t = 4.663, df = 31.624, p-value = 5.406e-05
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##   5.001707 12.767524
## sample estimates:
## mean of x mean of y 
##  20.88462  12.00000

T-test of Group 2 vs 3 (control vs mild)

# group 2 vs 3
t.test(protein_data_clean[protein_data_clean$Group==2,3], protein_data_clean[protein_data_clean$Group==3,3])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 2, 3] and protein_data_clean[protein_data_clean$Group == 3, 3]
## t = 0.79327, df = 28.744, p-value = 0.4341
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##  -1.513432  3.430098
## sample estimates:
## mean of x mean of y 
##  12.00000  11.04167

T-test of Group 1 vs 3 (control vs mild)

# group 1 vs 3
t.test(protein_data_clean[protein_data_clean$Group==1,3], protein_data_clean[protein_data_clean$Group==3,3])

## 
##  Welch Two Sample t-test
## 
## data:  protein_data_clean[protein_data_clean$Group == 1, 3] and protein_data_clean[protein_data_clean$Group == 3, 3]
## t = 4.9346, df = 38.265, p-value = 1.604e-05
## alternative hypothesis: true difference in means is not equal to 0
## 95 percent confidence interval:
##   5.80584 13.88006
## sample estimates:
## mean of x mean of y 
##  20.88462  11.04167

Summary of Days since onset

• Signiicantly More Males than females in 0 vs 3, 1 vs 3
• no significant differnce when comparing controls against all cases (grouped)

Correlation days since onset with severity

Any Correlation between the days since onset and protein expression?

First check correlation with disease severity as distribution of days since onset was skewed by disease severity

Spearmans used as severity is coded as 0,1,2,3 and is classed as ordinal value while protein expression is continous (non-ordinal).

days_since_onset_cor<-rcorr(as.numeric(protein_data_clean[,1]), as.numeric(protein_data_clean[,3]), type="spearman")
days_since_onset_cor$P[2]

## [1] 8.055056e-08

days_since_onset_cor$r[2]

## [1] 0.5373206

Days since onset is correlated with disease severity. However, this maybe driven by group 1.

correlation of days of onset within group 1

#create subset of group 1
group1_only<-(subset(protein_data_clean, protein_data_clean$Group==1))

# group1 exprs only
group1_exprs_only<-group1_only[10:ncol(group1_only)]

# empty dataframe for results
grp1_cor_days<-as.data.frame(matrix(nrow=nrow(assay_data_clean), ncol=2))

# colnames
colnames(grp1_cor_days)<-c("cor", "p-value")

# for each protein perform correlation analysis against disease severity (i.e 0,1,2,3) - use expression data only
for (x in 1:ncol(group1_exprs_only)) {
  # add protein name
  rownames(grp1_cor_days)[x]<-colnames(group1_exprs_only)[x]
  # perform test
  cor_results_temp<-rcorr(group1_only[,3], group1_exprs_only[,x], type="pearson")
  #extract cor
  grp1_cor_days[x,1]<-cor_results_temp$r[2]
  #extract p -value
  grp1_cor_days[x,2]<-cor_results_temp$P[2]
}

# check
head(grp1_cor_days)

##                   cor    p-value
## OID00379  0.326065145 0.10402926
## OID00380  0.328589705 0.10122181
## OID00381  0.024365172 0.90595237
## OID00382  0.457958610 0.01864101
## OID00383 -0.285302193 0.15772128
## OID00384 -0.003808262 0.98526918

# sort by p-value
grp1_cor_days<-grp1_cor_days[order(grp1_cor_days$`p-value`, -abs(-grp1_cor_days$cor)),]

# sort by pvalue
plot(grp1_cor_days[,1], grp1_cor_days[,2])

# how mnay sig
nrow(subset(grp1_cor_days, grp1_cor_days$`p-value`<0.05))

## [1] 51

# sort by pvalue
grp1_cor_days<-grp1_cor_days[order(grp1_cor_days$`p-value`, -abs(grp1_cor_days$cor)),]

#results table
datatable(grp1_cor_days)

how many postive/negative correlated

#positive
nrow(subset(grp1_cor_days, grp1_cor_days$`p-value`<=0.05 & grp1_cor_days$cor>0))

## [1] 24

#negative
nrow(subset(grp1_cor_days, grp1_cor_days$`p-value`<=0.05 & grp1_cor_days$cor<0))

## [1] 27

plot in group

ggplot(data=group1_only[c(3,(grep("OID01018", colnames(group1_only))))], 
       aes(x=Days.since.onset, y=OID01018)) +
  geom_point() +
  geom_smooth(method = "lm", se=F)

## `geom_smooth()` using formula 'y ~ x'

## Warning: Removed 1 rows containing non-finite values (stat_smooth).

## Warning: Removed 1 rows containing missing values (geom_point).

plot in all groups (minus control as they are zero)

# create cases only group
cases_only<-protein_data_clean[protein_data_clean$Group!=0,]

# plot
ggplot(data=cases_only[c(3,(grep("OID01018", colnames(cases_only))))], 
       aes(x=Days.since.onset, y=OID01018)) +
  geom_point() +
  geom_smooth(method = "lm", se=F)

## `geom_smooth()` using formula 'y ~ x'

## Warning: Removed 1 rows containing non-finite values (stat_smooth).

## Warning: Removed 1 rows containing missing values (geom_point).

correlation of days of onset within group 2

#create subset of group 2
group2_only<-(subset(protein_data_clean, protein_data_clean$Group==2))

# group2 exprs only
group2_exprs_only<-group2_only[10:ncol(group2_only)]

# empty dataframe for results
grp2_cor_days<-as.data.frame(matrix(nrow=nrow(assay_data_clean), ncol=2))

# colnames
colnames(grp2_cor_days)<-c("cor", "p-value")

# for each protein perform correlation analysis against disease severity (i.e 0,1,2,3) - use expression data only
for (x in 1:ncol(group2_exprs_only)) {
  # add protein name
  rownames(grp2_cor_days)[x]<-colnames(group2_exprs_only)[x]
  # perform test
  cor_results_temp<-rcorr(group2_only[,3], group2_exprs_only[,x], type="spearman")
  #extract cor
  grp2_cor_days[x,1]<-cor_results_temp$r[2]
  #extract p -value
  grp2_cor_days[x,2]<-cor_results_temp$P[2]
}

# check
head(grp2_cor_days)

##                 cor   p-value
## OID00379  0.3697610 0.3273605
## OID00380  0.5126231 0.1582071
## OID00381  0.5042195 0.1663127
## OID00382  0.2521097 0.5128372
## OID00383 -0.5378341 0.1352889
## OID00384  0.2689171 0.4841161

# sort by p-value
grp2_cor_days<-grp2_cor_days[order(grp2_cor_days$`p-value`, -abs(-grp2_cor_days$cor)),]

# sort by pvalue
plot(grp2_cor_days[,1], grp2_cor_days[,2])

# how mnay sig
nrow(subset(grp2_cor_days, grp2_cor_days$`p-value`<0.05))

## [1] 12

# sort by pvalue
grp2_cor_days<-grp2_cor_days[order(grp2_cor_days$`p-value`, -abs(grp2_cor_days$cor)),]

#results table
datatable(grp2_cor_days)

how many postive/negative correlated

#positive
nrow(subset(grp2_cor_days, grp2_cor_days$`p-value`<=0.05 & grp2_cor_days$cor>0))

## [1] 8

#negative
nrow(subset(grp2_cor_days, grp2_cor_days$`p-value`<=0.05 & grp2_cor_days$cor<0))

## [1] 4

plot

ggplot(data=group2_only[c(3,(grep("OID00968", colnames(group2_only))))], 
       aes(x=Days.since.onset, y=OID00968)) +
  geom_point() +
  geom_smooth(method = "lm", se=F)

## `geom_smooth()` using formula 'y ~ x'

plot in all cases

# plot
ggplot(data=cases_only[c(3,(grep("OID00968", colnames(cases_only))))], 
       aes(x=Days.since.onset, y=OID00968)) +
  geom_point() +
  geom_smooth(method = "lm", se=F)

## `geom_smooth()` using formula 'y ~ x'

## Warning: Removed 1 rows containing non-finite values (stat_smooth).

## Warning: Removed 1 rows containing missing values (geom_point).

correlation of days of onset within group 3

#create subset of group 3
group3_only<-(subset(protein_data_clean, protein_data_clean$Group==3))

# group3 exprs only
group3_exprs_only<-group3_only[10:ncol(group3_only)]

# empty dataframe for results
grp3_cor_days<-as.data.frame(matrix(nrow=nrow(assay_data_clean), ncol=2))

# colnames
colnames(grp3_cor_days)<-c("cor", "p-value")

# for each protein perform correlation analysis against disease severity (i.e 0,1,2,3) - use expression data only
for (x in 1:ncol(group3_exprs_only)) {
  # add protein name
  rownames(grp3_cor_days)[x]<-colnames(group3_exprs_only)[x]
  # perform test
  cor_results_temp<-rcorr(group3_only[,3], group3_exprs_only[,x], type="spearman")
  #extract cor
  grp3_cor_days[x,1]<-cor_results_temp$r[2]
  #extract p -value
  grp3_cor_days[x,2]<-cor_results_temp$P[2]
}

# check
head(grp3_cor_days)

##                  cor   p-value
## OID00379  0.30015918 0.1541338
## OID00380 -0.06213208 0.7730304
## OID00381  0.06957042 0.7466791
## OID00382 -0.15401754 0.4724157
## OID00383  0.25421645 0.2306313
## OID00384  0.08619732 0.6888009

# sort by p-value
grp3_cor_days<-grp3_cor_days[order(grp3_cor_days$`p-value`, -abs(-grp3_cor_days$cor)),]

# sort by pvalue
plot(grp3_cor_days[,1], grp3_cor_days[,2])

# how mnay sig
nrow(subset(grp3_cor_days, grp3_cor_days$`p-value`<0.05))

## [1] 49

# sort by pvalue
grp3_cor_days<-grp3_cor_days[order(grp3_cor_days$`p-value`, -abs(grp3_cor_days$cor)),]

#results table
datatable(grp3_cor_days)

how many postive/negative correlated

#positive
nrow(subset(grp3_cor_days, grp3_cor_days$`p-value`<=0.05 & grp3_cor_days$cor>0))

## [1] 23

#negative
nrow(subset(grp3_cor_days, grp3_cor_days$`p-value`<=0.05 & grp3_cor_days$cor<0))

## [1] 26

ggplot(data=group3_only[c(3,(grep("OID01021", colnames(group3_only))))], 
       aes(x=Days.since.onset, y=OID01021)) +
  geom_point() +
  geom_smooth(method = "lm", se=F)

## `geom_smooth()` using formula 'y ~ x'

plot in all cases

# plot
ggplot(data=cases_only[c(3,(grep("OID01021", colnames(cases_only))))], 
       aes(x=Days.since.onset, y=OID01021)) +
  geom_point() +
  geom_smooth(method = "lm", se=F)

## `geom_smooth()` using formula 'y ~ x'

## Warning: Removed 1 rows containing non-finite values (stat_smooth).

## Warning: Removed 1 rows containing missing values (geom_point).

Check overlap of correlated proteins

# move protein name column + add analysis
grp1_cor_days$OLINK.ID<-rownames(grp1_cor_days)
grp2_cor_days$OLINK.ID<-rownames(grp2_cor_days)
grp3_cor_days$OLINK.ID<-rownames(grp3_cor_days)

grp1_cor_days$analysis<-"grp1"
grp2_cor_days$analysis<-"grp2"
grp3_cor_days$analysis<-"grp3"

# merge
grp_cor_merged<-rbind((subset(grp1_cor_days, grp1_cor_days$`p-value`<0.05)),
(subset(grp2_cor_days, grp2_cor_days$`p-value`<0.05)),
(subset(grp3_cor_days, grp3_cor_days$`p-value`<0.05)))

# which grps are they
grp_cor_merged_frq<-(as.data.frame(table(grp_cor_merged$OLINK.ID)))
grp_cor_merged_frq<-grp_cor_merged_frq[grp_cor_merged_frq$Freq>1,]

# subset to duplicates
grp_cor_merged_dups<-subset(grp_cor_merged, grp_cor_merged$OLINK.ID %in% grp_cor_merged_frq$Var1)
grp_cor_merged_dups<-grp_cor_merged_dups[order(grp_cor_merged_dups$`p-value`),]
head(grp_cor_merged_dups)

##                  cor     p-value OLINK.ID analysis
## OID01018  -0.6158719 0.001047063 OID01018     grp1
## OID004511 -0.6121322 0.001477203 OID00451     grp3
## OID003401 -0.6112571 0.001507046 OID00340     grp3
## OID00528  -0.5738006 0.002177467 OID00528     grp1
## OID005281 -0.5941927 0.002201023 OID00528     grp3
## OID00340  -0.5801698 0.002364417 OID00340     grp1

table(grp_cor_merged_dups$OLINK.ID, grp_cor_merged_dups$analysis)

##           
##            grp1 grp2 grp3
##   OID00340    1    0    1
##   OID00351    1    0    1
##   OID00361    1    0    1
##   OID00362    0    1    1
##   OID00432    0    1    1
##   OID00451    1    0    1
##   OID00457    0    1    1
##   OID00528    1    0    1
##   OID00945    1    0    1
##   OID00993    1    0    1
##   OID01001    1    1    0
##   OID01018    1    0    1
##   OID01019    1    0    1

#get gene name
subset(assay_data_clean, assay_data_clean$OLINK.ID %in% rownames(grp_cor_merged_dups))

##                               Assay Gene.ID Uniprot.ID OLINK.ID      LOD
## 54  Olink CARDIOVASCULAR II(v.5006)    HO-1     P09601 OID00432  0.93744
## 73  Olink CARDIOVASCULAR II(v.5006)   FABP2     P12104 OID00451  1.04637
## 79  Olink CARDIOVASCULAR II(v.5006)    ACE2     Q9BYF1 OID00457  0.78739
## 102   Olink IMMUNE RESPONSE(v.3203)    IRF9     Q00978 OID00945  0.83330
## 150   Olink IMMUNE RESPONSE(v.3203)    IL10     P22301 OID00993  1.58493
## 158   Olink IMMUNE RESPONSE(v.3203)   FAM3B     P58499 OID01001  0.79853
## 175   Olink IMMUNE RESPONSE(v.3203)   DDX58     O95786 OID01018  0.79252
## 176   Olink IMMUNE RESPONSE(v.3203) IL12RB1     P42701 OID01019  1.12822
## 242      Olink INFLAMMATION(v.3022)    IL10     P22301 OID00528  1.48617
## 330         Olink NEUROLOGY(v.8012)   G-CSF     P09919 OID00340  1.34264
## 341         Olink NEUROLOGY(v.8012)   BMP-4     P12644 OID00351  1.72820
## 351         Olink NEUROLOGY(v.8012) N-CDase     Q9NR71 OID00361 -1.12374
## 352         Olink NEUROLOGY(v.8012)    NAAA     Q02083 OID00362  0.72062

Plot most sig protein in more than 1 group

ggplot(data=cases_only[c(3,(grep("OID01018", colnames(cases_only))))], 
       aes(x=Days.since.onset, y=OID01018)) +
  geom_point() +
  geom_smooth(method = "lm", se=F)

## `geom_smooth()` using formula 'y ~ x'

## Warning: Removed 1 rows containing non-finite values (stat_smooth).

## Warning: Removed 1 rows containing missing values (geom_point).

Summary of days since onset correlation

• 42 proteins significantly correlated with days in group 1
• 12 proteins significantly correlated with days in group 2
• 49 proteins significantly correlated with days in group 3
• 13 proteins in more in 2 groups are correlated with days

days since onset - correlation with controls

MISLEADINg AS SEVERE AND MILD ARE DIFFERENT END OF THE DAYS.SNCE.ONSET SEPCTRUM. any protein that will be DE will alao be correlated. Better to use within groups to see effect of days on expression of protein.

Includes controls which are set to 0.

# copy dataset and change to case vs control
control_vs_case_dataset<-protein_data_clean
# change to case and control
control_vs_case_dataset$Group<-ifelse(control_vs_case_dataset$Group==0, 0, 1)

# exprs only
control_vs_case_dataset_exprs_onlly<-control_vs_case_dataset[10:ncol(control_vs_case_dataset)]

# empty dataframe for results
control_vs_case_cor_days<-as.data.frame(matrix(nrow=nrow(assay_data_clean), ncol=2))

# colnames
colnames(control_vs_case_cor_days)<-c("cor", "p-value")

# for each protein perform correlation analysis against disease severity (i.e 0,1,2,3) - use expression data only
for (x in 1:ncol(control_vs_case_dataset_exprs_onlly)) {
  # add protein name
  rownames(control_vs_case_cor_days)[x]<-colnames(control_vs_case_dataset_exprs_onlly)[x]
  # perform test
  cor_results_temp<-rcorr(control_vs_case_dataset[,3], control_vs_case_dataset_exprs_onlly[,x], type="spearman")
  #extract cor
  control_vs_case_cor_days[x,1]<-cor_results_temp$r[2]
  #extract p -value
  control_vs_case_cor_days[x,2]<-cor_results_temp$P[2]
}

# check
head(control_vs_case_cor_days)

##                 cor      p-value
## OID00379  0.1190516 2.720778e-01
## OID00380 -0.3393009 1.304664e-03
## OID00381  0.2374262 2.680820e-02
## OID00382 -0.5174981 2.857371e-07
## OID00383  0.1888725 7.977194e-02
## OID00384 -0.0513918 6.364057e-01

# sort by p-value
control_vs_case_cor_days<-control_vs_case_cor_days[order(control_vs_case_cor_days$`p-value`, -abs(-control_vs_case_cor_days$cor)),]

# sort by pvalue
plot(control_vs_case_cor_days[,1], control_vs_case_cor_days[,2])

# how mnay sig
nrow(subset(control_vs_case_cor_days, control_vs_case_cor_days$`p-value`<0.05))

## [1] 151

# sort by pvalue
control_vs_case_cor_days<-control_vs_case_cor_days[order(control_vs_case_cor_days$`p-value`, -abs(control_vs_case_cor_days$cor)),]

#results table
datatable(control_vs_case_cor_days)

plot most sig

ggplot(data=control_vs_case_dataset[c(3,(grep("OID00550", colnames(control_vs_case_dataset))))], 
       aes(x=Days.since.onset, y=OID00550)) +
  geom_point() +
  geom_smooth(method = "lm", se=F)

## `geom_smooth()` using formula 'y ~ x'

days since onset - correlation without controls

controls removed.

# case_expr
case_expr_only<-cases_only[10:ncol(cases_only)]

# empty dataframe for results
ind_group_cor_severity<-as.data.frame(matrix(nrow=ncol(case_expr_only), ncol=2))

# colnames
colnames(ind_group_cor_severity)<-c("cor", "p-value")

# for each protein perform correlation analysis against disease severity (i.e 0,1,2,3) - use expression data only
for (x in 1:ncol(case_expr_only)) {
  # add protein name
  rownames(ind_group_cor_severity)[x]<-colnames(case_expr_only)[x]
  # perform test
  cor_results_temp<-rcorr(as.numeric(cases_only[,3]), case_expr_only[,x], type="spearman")
  #extract cor
  ind_group_cor_severity[x,1]<-cor_results_temp$r[2]
  #extract p -value
  ind_group_cor_severity[x,2]<-cor_results_temp$P[2]
}

# check
head(ind_group_cor_severity)

##                 cor      p-value
## OID00379  0.0428449 0.7472977676
## OID00380  0.1210376 0.3611486301
## OID00381 -0.3495242 0.0066583547
## OID00382  0.1352411 0.3071209932
## OID00383 -0.4571489 0.0002727858
## OID00384 -0.3465956 0.0071623229

# sort by p-value
ind_group_cor_severity<-ind_group_cor_severity[order(ind_group_cor_severity$`p-value`, -abs(-ind_group_cor_severity$cor)),]

# sort by pvalue
plot(ind_group_cor_severity[,1], ind_group_cor_severity[,2])

# how mnay sig
nrow(subset(ind_group_cor_severity, ind_group_cor_severity$`p-value`<0.05))

## [1] 177

# merge with protein ID
ind_group_cor_severity<-merge(protein_mapping, ind_group_cor_severity, by="row.names")
# move rownames
rownames(ind_group_cor_severity)<-ind_group_cor_severity$Row.names
ind_group_cor_severity$Row.names<-NULL

# sort by pvalue
ind_group_cor_severity<-ind_group_cor_severity[order(ind_group_cor_severity$`p-value`, -abs(ind_group_cor_severity$cor)),]

#results table
datatable(ind_group_cor_severity)

0 vs 1 Differential expression

Group 0 vs 1 (control vs mild)

# run DE
group_0_vs_1<-run_limma(groups=c(0,1))

## 
##  0  1 
## 28 26 
##                   Contrasts
## Levels             Diff
##   case               -1
##   control             1
##   age                 0
##   Gender              0
##   Days.since.onset    0
##        Diff
## Down    105
## NotSig  208
## Up       42

# number of DE proteins
nrow(subset(group_0_vs_1, group_0_vs_1$adj.P.Val<0.05))

## [1] 147

# merge with protein ID
group_0_vs_1<-merge_protin_ID(group_0_vs_1)

# results
datatable(format(group_0_vs_1, digits=5))

significant (adjusted p < 0.05) proteins belong to assay:

table(assay_data_clean[assay_data_clean$OLINK.ID %in% rownames(subset(group_0_vs_1, group_0_vs_1$adj.P.Val<0.05)),][1])

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                              32                              54 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                              31                              30

Volcano plot of DE results

EnhancedVolcano(group_0_vs_1,
                lab = rownames(group_0_vs_1),
                x = 'logFC',
                y = 'adj.P.Val')

Boxplot of most significant protein

ggplot(data=protein_data_clean[c(1,(grep("OID00981", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00981, fill=as.character(Group))) +
  geom_boxplot()

## Warning: Removed 1 rows containing non-finite values (stat_boxplot).

0 vs 2 Differential expression

Group 0 vs 2 (control vs mild)

# run DE
group_0_vs_2<-run_limma(groups=c(0,2))

## 
##  0  2 
## 28  9 
##                   Contrasts
## Levels             Diff
##   case               -1
##   control             1
##   age                 0
##   Gender              0
##   Days.since.onset    0
##        Diff
## Down     57
## NotSig  290
## Up        8

# number of DE proteins
nrow(subset(group_0_vs_2, group_0_vs_2$adj.P.Val<0.05))

## [1] 65

# merge with protein ID
group_0_vs_2<-merge_protin_ID(group_0_vs_2)

# results
datatable(format(group_0_vs_2, digits=5))

significant (adjusted p < 0.05) proteins belong to assay:

table(assay_data_clean[assay_data_clean$OLINK.ID %in% rownames(subset(group_0_vs_2, group_0_vs_2$adj.P.Val<0.05)),][1])

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                              17                              28 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                               7                              13

Volcano plot of DE results

EnhancedVolcano(group_0_vs_2,
                lab = rownames(group_0_vs_2),
                x = 'logFC',
                y = 'adj.P.Val')

Boxplot of most significant protein

ggplot(data=protein_data_clean[c(1,(grep("OID00374", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00374, fill=as.character(Group))) +
  geom_boxplot()

## Warning: Removed 1 rows containing non-finite values (stat_boxplot).

0 vs 3 Differential expression

Group 0 vs 3 (control vs mild)

# run DE
group_0_vs_3<-run_limma(groups=c(0,3))

## 
##  0  3 
## 28 24 
##                   Contrasts
## Levels             Diff
##   case               -1
##   control             1
##   age                 0
##   Gender              0
##   Days.since.onset    0
##        Diff
## Down    105
## NotSig  168
## Up       82

# number of DE proteins
nrow(subset(group_0_vs_3, group_0_vs_3$adj.P.Val<0.05))

## [1] 187

# merge with protein ID
group_0_vs_3<-merge_protin_ID(group_0_vs_3)

# results
datatable(format(group_0_vs_3, digits=5))

significant (adjusted p < 0.05) proteins belong to assay:

table(assay_data_clean[assay_data_clean$OLINK.ID %in% rownames(subset(group_0_vs_3, group_0_vs_3$adj.P.Val<0.05)),][1])

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                              41                              58 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                              44                              44

Volcano plot of DE results

EnhancedVolcano(group_0_vs_3,
                lab = rownames(group_0_vs_3),
                x = 'logFC',
                y = 'adj.P.Val')

Most significant protein is NF2 again.

1 vs 2 Differential expression

Group 1 vs 2 (mild vs moderate)

# run DE
group_1_vs_2<-run_limma(groups=c(1,2))

## 
##  1  2 
## 26  9 
##                   Contrasts
## Levels             Diff
##   case               -1
##   control             1
##   age                 0
##   Gender              0
##   Days.since.onset    0
##        Diff
## Down     49
## NotSig  209
## Up       97

# number of DE proteins
nrow(subset(group_1_vs_2, group_1_vs_2$adj.P.Val<0.05))

## [1] 146

# merge with protein ID
group_1_vs_2<-merge_protin_ID(group_1_vs_2)

# results
datatable(format(group_1_vs_2, digits=5))

significant (adjusted p < 0.05) proteins belong to assay:

table(assay_data_clean[assay_data_clean$OLINK.ID %in% rownames(subset(group_1_vs_2, group_1_vs_2$adj.P.Val<0.05)),][1])

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                              34                              39 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                              37                              36

Volcano plot of DE results

EnhancedVolcano(group_1_vs_2,
                lab = rownames(group_1_vs_2),
                x = 'logFC',
                y = 'adj.P.Val')

Plot most sig protein

# plot all groups most correlated
ggplot(data=protein_data_clean[c(1,(grep("OID00541", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00541 , fill=as.character(Group))) +
  geom_boxplot()

1 vs 3 Differential expression

Group 1 vs 3 (mild vs severe)

# run DE
group_1_vs_3<-run_limma(groups=c(1,3))

## 
##  1  3 
## 26 24 
##                   Contrasts
## Levels             Diff
##   case               -1
##   control             1
##   age                 0
##   Gender              0
##   Days.since.onset    0
##        Diff
## Down     39
## NotSig  158
## Up      158

# number of DE proteins
nrow(subset(group_1_vs_3, group_1_vs_3$adj.P.Val<0.05))

## [1] 197

# merge with protein ID
group_1_vs_3<-merge_protin_ID(group_1_vs_3)

# results
datatable(format(group_1_vs_3, digits=5))

significant (adjusted p < 0.05) proteins belong to assay:

table(assay_data_clean[assay_data_clean$OLINK.ID %in% rownames(subset(group_1_vs_3, group_1_vs_3$adj.P.Val<0.05)),][1])

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                              53                              48 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                              48                              48

Volcano plot of DE results

EnhancedVolcano(group_1_vs_3,
                lab = rownames(group_1_vs_3),
                x = 'logFC',
                y = 'adj.P.Val')

Plot most sig protein

# plot all groups most correlated
ggplot(data=protein_data_clean[c(1,(grep("OID00459", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00459 , fill=as.character(Group))) +
  geom_boxplot()

2 vs 3 Differential expression

Group 2 vs 3 (moderate vs severe)

# run DE
group_2_vs_3<-run_limma(groups=c(2,3))

## 
##  2  3 
##  9 24 
##                   Contrasts
## Levels             Diff
##   case               -1
##   control             1
##   age                 0
##   Gender              0
##   Days.since.onset    0
##        Diff
## Down      1
## NotSig  291
## Up       63

# number of DE proteins
nrow(subset(group_2_vs_3, group_2_vs_3$adj.P.Val<0.05))

## [1] 64

# merge with protein ID
group_2_vs_3<-merge_protin_ID(group_2_vs_3)

# results
datatable(format(group_2_vs_3, digits=5))

significant (adjusted p < 0.05) proteins belong to assay:

table(assay_data_clean[assay_data_clean$OLINK.ID %in% rownames(subset(group_2_vs_3, group_2_vs_3$adj.P.Val<0.05)),][1])

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                              17                              11 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                              15                              21

Volcano plot of DE results

EnhancedVolcano(group_2_vs_3,
                lab = rownames(group_2_vs_3),
                x = 'logFC',
                y = 'adj.P.Val')

Plot most sig protein

# plot all groups most correlated
ggplot(data=protein_data_clean[c(1,(grep("OID00517", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00517 , fill=as.character(Group))) +
  geom_boxplot()

case vs control

recode all cases (1,2,3) to case and compare to control

# copy dataset and change to case vs control
control_vs_case_dataset<-protein_data_clean
# change to case and control
control_vs_case_dataset$Group<-ifelse(control_vs_case_dataset$Group==0, 0, 1)
# check 
table(control_vs_case_dataset$Group)

## 
##  0  1 
## 28 59

# design a model - control will always be on top as 0
control_vs_case_design<-model.matrix(~0 + as.factor(control_vs_case_dataset$Group) + control_vs_case_dataset$Age + as.factor(control_vs_case_dataset$Gender) + control_vs_case_dataset$Days.since.onset)
colnames(control_vs_case_design)<-c("case", "control", "age", "Gender", "Days.since.onset")
# make contrast - what to compare
control_vs_case_contrast<- makeContrasts(Diff = control - case, levels=control_vs_case_design)
head(control_vs_case_contrast)

##                   Contrasts
## Levels             Diff
##   case               -1
##   control             1
##   age                 0
##   Gender              0
##   Days.since.onset    0

# apply linear model to each protein
# Robust regression provides an alternative to least squares regression that works with less restrictive assumptions. Specifically, it provides much better regression coefficient estimates when outliers are present in the data
control_vs_case_fit<-lmFit(t(control_vs_case_dataset[10:ncol(control_vs_case_dataset)]), design=control_vs_case_design,  method="robust", maxit=100)
# apply contrast
control_vs_case_contrast_fit<-contrasts.fit(control_vs_case_fit, control_vs_case_contrast)
# apply empirical Bayes smoothing to the SE
control_vs_case_ebays_fit<-eBayes(control_vs_case_contrast_fit)
# summary
print(summary(decideTests(control_vs_case_ebays_fit)))

##        Diff
## Down    149
## NotSig   86
## Up      120

# extract DE results
control_vs_case_DE_results<-topTable(control_vs_case_ebays_fit, n=ncol(control_vs_case_dataset), adjust.method="fdr", confint=TRUE)

# number of DE proteins
nrow(subset(control_vs_case_DE_results, control_vs_case_DE_results$adj.P.Val<0.05))

## [1] 269

# merge with protein ID
control_vs_case_DE_results<-merge_protin_ID(control_vs_case_DE_results)

# results
datatable(format(control_vs_case_DE_results, digits=5))

significant (adjusted p < 0.05) proteins belong to assay:

table(assay_data_clean[assay_data_clean$OLINK.ID %in% rownames(subset(control_vs_case_DE_results, control_vs_case_DE_results$adj.P.Val<0.05)),][1])

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                              65                              71 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                              62                              71

Volcano plot of DE results

EnhancedVolcano(control_vs_case_DE_results,
                lab = rownames(control_vs_case_DE_results),
                x = 'logFC',
                y = 'adj.P.Val')

NF2 most sig again

Longitudinal group 1

Differential expression between the 12 samples in group 1 that were repeated at different time points

• 6 patients = 12 samples
• baseline and first repeat assigned to samples
• controlling for age + days.since.onset + gender
• paired t-test

# extract group 1
group1_duplicates_data<-duplicate_samples_after_qc[duplicate_samples_after_qc$Group==1,]
# add column for longitudinal analysis - sample.repeat - baseline/first (baseline is always 1st sample)
group1_duplicates_data$sample.repeat<-"baseline"
group1_duplicates_data$sample.repeat[c(2,4,6,8,10,12)]<-"first"
# move sample.repeat to front
ncol(group1_duplicates_data)

## [1] 365

group1_duplicates_data<-group1_duplicates_data[c(365,1:364)]
# check
head(group1_duplicates_data, 12)[1:5]

##          sample.repeat Group Pat.Code Days.since.onset Gender
## sample1       baseline     1      101                8      M
## sample2          first     1      101               27      M
## sample6       baseline     1      105               17      F
## sample7          first     1      105               27      F
## sample13      baseline     1      111                8      F
## sample14         first     1      111               22      F
## sample15      baseline     1      112                7      M
## sample16         first     1      112               26      M
## sample19      baseline     1      115                2      M
## sample20         first     1      115               22      M
## sample24      baseline     1      119                9      F
## sample25         first     1      119               24      F

table(group1_duplicates_data$sample.repeat)

## 
## baseline    first 
##        6        6

# design a model - control will always be on top as 0
group1_longitudinal_design<-model.matrix(~0 + as.factor(group1_duplicates_data$sample.repeat) + group1_duplicates_data$Age + group1_duplicates_data$Days.since.onset + as.factor(group1_duplicates_data$Gender))
colnames(group1_longitudinal_design)<-c("baseline", "first", "age", "days.since.onset", "Gender")
# estimtate correlation between repeated patients
group1_corfit <- duplicateCorrelation(t(group1_duplicates_data[10:ncol(group1_duplicates_data)]),group1_longitudinal_design,block=group1_duplicates_data$Pat.Code)
group1_corfit$consensus

## [1] 0.5824546

# make contrast - what to compare
group1_longitudinal_contrast<- makeContrasts(Diff = first - baseline, levels=group1_longitudinal_design)
head(group1_longitudinal_contrast)

##                   Contrasts
## Levels             Diff
##   baseline           -1
##   first               1
##   age                 0
##   days.since.onset    0
##   Gender              0

# add inter-subject correlation to model
# apply linear model to each protein
# Robust regression provides an alternative to least squares regression that works with less restrictive assumptions. Specifically, it provides much better regression coefficient estimates when outliers are present in the data
group1_longitudinal_fit<-lmFit(t(group1_duplicates_data[10:ncol(group1_duplicates_data)]), design=group1_longitudinal_design,  method="robust", maxit=100, correlation = group1_corfit$consensus)
# apply contrast
group1_longitudinal_contrast_fit<-contrasts.fit(group1_longitudinal_fit, group1_longitudinal_contrast)
# apply empirical Bayes smoothing to the SE
group1_longitudinal_ebays_fit<-eBayes(group1_longitudinal_contrast_fit)

## Warning: Zero sample variances detected, have been offset away from zero

# summary
print(summary(decideTests(group1_longitudinal_ebays_fit)))

##        Diff
## Down      3
## NotSig  343
## Up       10

# extract DE results
group1_longitudinal_DE_results<-topTable(group1_longitudinal_ebays_fit, n=ncol(group1_duplicates_data), adjust.method="fdr", confint=TRUE)

# number of DE proteins
nrow(subset(group1_longitudinal_DE_results, group1_longitudinal_DE_results$adj.P.Val<0.05))

## [1] 13

# merge with protein ID
group1_longitudinal_DE_results<-merge_protin_ID(group1_longitudinal_DE_results)

# results
datatable(format(group1_longitudinal_DE_results, digits=5))

significant (adjusted p < 0.05) proteins belong to assay:

table(assay_data_clean[assay_data_clean$OLINK.ID %in% rownames(subset(group1_longitudinal_DE_results, group1_longitudinal_DE_results$adj.P.Val<0.05)),][1])

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                               5                               4 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                               2                               2

Boxplots of most significant proteins

# plot by group 1 duplicate
ggplot(data=group1_duplicates_data[c(1,(grep("OID00386", colnames(group1_duplicates_data))))], 
       aes(x=sample.repeat, y=OID00386, fill=sample.repeat)) +
  geom_boxplot()

Longitudinal group 3

Differential expression between the 10 samples in group 3 that were repeated at different time points

• 5 patients = 10 samples
• baseline and first repeat assigned to samples
• controlling for age + days.since.onset
• paired t-test
• All males (gender not modelled)

# extract group 3
group3_duplicates_data<-duplicate_samples_after_qc[duplicate_samples_after_qc$Group==3,]
# add column for longitudinal analysis - sample.repeat - baseline/first (baseline is always 1st sample)
group3_duplicates_data$sample.repeat<-"baseline"
group3_duplicates_data$sample.repeat[c(2,4,6,8,10)]<-"first"
# move sample.repeat to front
ncol(group3_duplicates_data)

## [1] 365

group3_duplicates_data<-group3_duplicates_data[c(365,1:364)]
# check
head(group3_duplicates_data, 12)[1:5]

##          sample.repeat Group Pat.Code Days.since.onset Gender
## sample39      baseline     3      304               11      M
## sample40         first     3      304               15      M
## sample43      baseline     3      307                6      M
## sample44         first     3      307                8      M
## sample47      baseline     3      310               10      M
## sample48         first     3      310               12      M
## sample54      baseline     3      315                9      M
## sample55         first     3      315               13      M
## sample58      baseline     3      318                6      M
## sample59         first     3      318                8      M

table(group3_duplicates_data$sample.repeat)

## 
## baseline    first 
##        5        5

# design a model - control will always be on top as 0
group3_longitudinal_design<-model.matrix(~0 + as.factor(group3_duplicates_data$sample.repeat) + group3_duplicates_data$Age + group3_duplicates_data$Days.since.onset)
colnames(group3_longitudinal_design)<-c("baseline", "first", "age", "days.since.onset")
# estimtate correlation between repeated patients
group3_corfit <- duplicateCorrelation(t(group3_duplicates_data[10:ncol(group3_duplicates_data)]),group3_longitudinal_design,block=group3_duplicates_data$Pat.Code)

## Warning in glmgam.fit(dx, dy, coef.start = start, tol = tol, maxit = maxit, :
## Too much damping - convergence tolerance not achievable

## Warning in glmgam.fit(dx, dy, coef.start = start, tol = tol, maxit = maxit, :
## Too much damping - convergence tolerance not achievable

group3_corfit$consensus

## [1] 0.42842

# make contrast - what to compare
group3_longitudinal_contrast<- makeContrasts(Diff = first - baseline, levels=group3_longitudinal_design)
head(group3_longitudinal_contrast)

##                   Contrasts
## Levels             Diff
##   baseline           -1
##   first               1
##   age                 0
##   days.since.onset    0

# add inter-subject correlation to model
# apply linear model to each protein
# Robust regression provides an alternative to least squares regression that works with less restrictive assumptions. Specifically, it provides much better regression coefficient estimates when outliers are present in the data
group3_longitudinal_fit<-lmFit(t(group3_duplicates_data[10:ncol(group3_duplicates_data)]), design=group3_longitudinal_design,  method="robust", maxit=1000, correlation = group3_corfit$consensus)
# apply contrast
group3_longitudinal_contrast_fit<-contrasts.fit(group3_longitudinal_fit, group3_longitudinal_contrast)
# apply empirical Bayes smoothing to the SE
group3_longitudinal_ebays_fit<-eBayes(group3_longitudinal_contrast_fit)

## Warning: Zero sample variances detected, have been offset away from zero

# summary
print(summary(decideTests(group3_longitudinal_ebays_fit)))

##        Diff
## Down      3
## NotSig  350
## Up        3

# extract DE results
group3_longitudinal_DE_results<-topTable(group3_longitudinal_ebays_fit, n=ncol(group3_duplicates_data), adjust.method="fdr", confint=TRUE)

# number of DE proteins
nrow(subset(group3_longitudinal_DE_results, group3_longitudinal_DE_results$adj.P.Val<0.05))

## [1] 6

# merge with protein ID
group3_longitudinal_DE_results<-merge_protin_ID(group3_longitudinal_DE_results)

# results
datatable(format(group3_longitudinal_DE_results, digits=5))

significant (adjusted p < 0.05) proteins belong to assay:

table(assay_data_clean[assay_data_clean$OLINK.ID %in% rownames(subset(group3_longitudinal_DE_results, group3_longitudinal_DE_results$adj.P.Val<0.05)),][1])

## 
## Olink CARDIOVASCULAR II(v.5006)   Olink IMMUNE RESPONSE(v.3203) 
##                               2                               1 
##      Olink INFLAMMATION(v.3022)         Olink NEUROLOGY(v.8012) 
##                               0                               3

Boxplot of most sig

# plot by group 3 duplicate
ggplot(data=group3_duplicates_data[c(1,(grep("OID00424", colnames(group3_duplicates_data))))], 
       aes(x=sample.repeat, y=OID00424 , fill=sample.repeat)) +
  geom_boxplot()

correlation analysis of severity

Any Correlation between the severity of symptoms and protein expression?

from control->mild->moderate->severe

# empty dataframe for results
cor_severity<-as.data.frame(matrix(nrow=nrow(assay_data_clean), ncol=2))

# colnames
colnames(cor_severity)<-c("cor", "p-value")

# for each protein perform correlation analysis against disease severity (i.e 0,1,2,3) - use expression data only
for (x in 1:ncol(exprs_only)) {
  # add protein name
  rownames(cor_severity)[x]<-colnames(exprs_only)[x]
  # perform test
  cor_results_temp<-rcorr(as.numeric(protein_data_clean[,1]), exprs_only[,x], type="spearman")
  #extract cor
  cor_severity[x,1]<-cor_results_temp$r[2]
  #extract p -value
  cor_severity[x,2]<-cor_results_temp$P[2]
}

# check
head(cor_severity)

##                 cor      p-value
## OID00379  0.2246237 3.646928e-02
## OID00380 -0.3972408 1.391701e-04
## OID00381  0.6636468 2.464917e-12
## OID00382 -0.5503791 3.342356e-08
## OID00383  0.6658724 1.961320e-12
## OID00384  0.3863220 2.191736e-04

# sort by p-value
cor_severity<-cor_severity[order(cor_severity$`p-value`, -abs(-cor_severity$cor)),]

# sort by pvalue
plot(cor_severity[,1], cor_severity[,2])

# how mnay sig
nrow(subset(cor_severity, cor_severity$`p-value`<0.05))

## [1] 286

# merge with protein ID
cor_severity<-merge(protein_mapping, cor_severity, by="row.names")
# move rownames
rownames(cor_severity)<-cor_severity$Row.names
cor_severity$Row.names<-NULL

# sort by pvalue
cor_severity<-cor_severity[order(cor_severity$`p-value`, -abs(cor_severity$cor)),]

#results table
datatable(cor_severity)

Plot most correlated

# plot all groups most correlated
ggplot(data=protein_data_clean[c(1,(grep("OID00999", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00999 , fill=as.character(Group))) +
  geom_boxplot()

## Warning: Removed 1 rows containing non-finite values (stat_boxplot).

Plot most anti-correlated

# plot most anti-correlated
ggplot(data=protein_data_clean[c(1,(grep("OID00951", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00951, fill=as.character(Group))) +
  geom_boxplot()

## Warning: Removed 1 rows containing non-finite values (stat_boxplot).

Apply function to Tau

#Tau
Tau_cor<-run_simoa_cor("Tau")
# how many sig
nrow(subset(Tau_cor, Tau_cor$`p-value`<=0.05))

## [1] 97

# head
head(Tau_cor)

##                   Assay        Gene.ID Uniprot.ID        cor      p-value
## OID00380 Cardiovascular         ANGPT1     Q15389 -0.5989159 8.915899e-10
## OID00401 Cardiovascular PDGF subunit B     P01127 -0.4609265 7.033602e-06
## OID00382 Cardiovascular         CD40-L     P29965 -0.4245190 4.167934e-05
## OID00449 Cardiovascular         HB-EGF     Q99075 -0.4218897 4.703018e-05
## OID00439 Cardiovascular          CCL17     Q92583 -0.4210625 4.884150e-05
## OID00300      Neurology         SCARB2     Q14108  0.3940164 1.743611e-04

Plot most correlated

# plot all groups most correlated
ggplot(data=protein_data_clean[c(1,(grep("OID00380", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00380 , fill=as.character(Group))) +
  geom_boxplot()

Apply function to NfL

#NfL
NfL_cor<-run_simoa_cor("NfL")
# how many sig
nrow(subset(NfL_cor, NfL_cor$`p-value`<=0.05))

## [1] 233

# head
head(NfL_cor)

##                   Assay   Gene.ID Uniprot.ID       cor      p-value
## OID00370      Neurology     EDA2R     Q9HAV5 0.6617428 4.014344e-12
## OID00394 Cardiovascular TNFRSF11A     Q9Y6Q6 0.6534596 6.847634e-12
## OID00396 Cardiovascular  TRAIL-R2     O14763 0.6529903 7.170931e-12
## OID00346      Neurology      SKR3     P37023 0.6454287 1.963896e-11
## OID00479   Inflammation       OPG     O00300 0.6402975 2.423617e-11
## OID00326      Neurology      LAYN     Q6UX15 0.6422911 2.636380e-11

Plot most correlated

# plot all groups most correlated
ggplot(data=protein_data_clean[c(1,(grep("OID00370", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00370 , fill=as.character(Group))) +
  geom_boxplot()

## Warning: Removed 1 rows containing non-finite values (stat_boxplot).

Apply function to GFAP

#GFAP
GFAP_cor<-run_simoa_cor("GFAP")
# how many sig
nrow(subset(GFAP_cor, GFAP_cor$`p-value`<=0.05))

## [1] 165

# head
head(GFAP_cor)

##                   Assay   Gene.ID Uniprot.ID       cor      p-value
## OID00455 Cardiovascular       BNP     P16860 0.5085301 4.937307e-07
## OID00394 Cardiovascular TNFRSF11A     Q9Y6Q6 0.4748341 3.371737e-06
## OID00552   Inflammation    CX3CL1     P78423 0.4748163 3.374980e-06
## OID01018         Immune     DDX58     O95786 0.4678390 5.574783e-06
## OID00300      Neurology    SCARB2     Q14108 0.4649629 6.474506e-06
## OID00399 Cardiovascular        TF     P13726 0.4623677 6.527273e-06

Plot most correlated

# plot all groups most correlated
ggplot(data=protein_data_clean[c(1,(grep("OID00380", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00380 , fill=as.character(Group))) +
  geom_boxplot()

Clean expression data

saveRDS(protein_data_clean, file=paste(output_dir, "Processed_data/protein_data_clean.RDS", sep="/"))

Clean protein info

saveRDS(assay_data_clean, file=paste(output_dir, "Processed_data/assay_data_clean.RDS", sep="/"))

0 vs 1

saveRDS(group_0_vs_1, file=paste(output_dir, "Result/group_0_vs_1.RDS", sep="/"))
write.table(subset(group_0_vs_1, group_0_vs_1$adj.P.Val<=0.05)[3], file="Result/group_0_vs_1.txt", col.names = F, row.names = F, quote=F)

0 vs 2

saveRDS(group_0_vs_2, file=paste(output_dir, "Result/group_0_vs_2.RDS", sep="/"))
write.table(subset(group_0_vs_2, group_0_vs_2$adj.P.Val<=0.05)[3], file="Result/group_0_vs_2.txt", col.names = F, row.names = F, quote=F)

0 vs 3

saveRDS(group_0_vs_3, file=paste(output_dir, "Result/group_0_vs_3.RDS", sep="/"))
write.table(subset(group_0_vs_3, group_0_vs_3$adj.P.Val<=0.05)[3], file="Result/group_0_vs_3.txt", col.names = F, row.names = F, quote=F)

1 vs 2

saveRDS(group_1_vs_2, file=paste(output_dir, "Result/group_1_vs_2.RDS", sep="/"))
write.table(subset(group_1_vs_2, group_1_vs_2$adj.P.Val<=0.05)[3], file="Result/group_1_vs_2.txt", col.names = F, row.names = F, quote=F)

1 vs 3

saveRDS(group_1_vs_3, file=paste(output_dir, "Result/group_1_vs_3.RDS", sep="/"))
write.table(subset(group_1_vs_3, group_1_vs_3$adj.P.Val<=0.05)[3], file="Result/group_1_vs_3.txt", col.names = F, row.names = F, quote=F)

2 vs 3

saveRDS(group_2_vs_3, file=paste(output_dir, "Result/group_2_vs_3.RDS", sep="/"))
write.table(subset(group_2_vs_3, group_2_vs_3$adj.P.Val<=0.05)[3], file="Result/group_2_vs_3.txt", col.names = F, row.names = F, quote=F)

control vs case

saveRDS(control_vs_case_DE_results, file=paste(output_dir, "Result/control_vs_case_DE_results.RDS", sep="/"))
write.table(subset(control_vs_case_DE_results, control_vs_case_DE_results$adj.P.Val<=0.05)[3], file="Result/control_vs_case_DE_results.txt", col.names = F, row.names = F, quote=F)

Longitudinal group 1

saveRDS(group1_longitudinal_DE_results, file=paste(output_dir, "Result/group1_longitudinal_DE_results.RDS", sep="/"))
write.table(subset(group1_longitudinal_DE_results, group1_longitudinal_DE_results$adj.P.Val<=0.05)[3], file="Result/group1_longitudinal_DE_results.txt", col.names = F, row.names = F, quote=F)

Longitudinal group 3

saveRDS(group3_longitudinal_DE_results, file=paste(output_dir, "Result/group3_longitudinal_DE_results.RDS", sep="/"))
write.table(subset(group3_longitudinal_DE_results, group3_longitudinal_DE_results$adj.P.Val<=0.05)[3], file="Result/group3_longitudinal_DE_results.txt", col.names = F, row.names = F, quote=F)

Infection severity correlation

saveRDS(cor_severity, file=paste(output_dir, "Result/cor_severity.RDS", sep="/"))

SIMOA protein cor

write.table(GFAP_cor, file=paste(output_dir, "Result/GFAP_correlation.txt", sep="/"), quote=F, sep="\t")
write.table(Tau_cor, file=paste(output_dir, "Result/Tau_correlation.txt", sep="/"), quote=F, sep="\t")
write.table(NfL_cor, file=paste(output_dir, "Result/NfL_correlation.txt", sep="/"), quote=F, sep="\t")

Merge results for Shiny app

# check colnames

colnames(group_0_vs_1)

##  [1] "Assay"      "Gene.ID"    "Uniprot.ID" "logFC"      "CI.L"      
##  [6] "CI.R"       "AveExpr"    "t"          "P.Value"    "adj.P.Val" 
## [11] "B"

colnames(group_0_vs_2)

##  [1] "Assay"      "Gene.ID"    "Uniprot.ID" "logFC"      "CI.L"      
##  [6] "CI.R"       "AveExpr"    "t"          "P.Value"    "adj.P.Val" 
## [11] "B"

colnames(group_0_vs_3)

##  [1] "Assay"      "Gene.ID"    "Uniprot.ID" "logFC"      "CI.L"      
##  [6] "CI.R"       "AveExpr"    "t"          "P.Value"    "adj.P.Val" 
## [11] "B"

colnames(group_1_vs_2)

##  [1] "Assay"      "Gene.ID"    "Uniprot.ID" "logFC"      "CI.L"      
##  [6] "CI.R"       "AveExpr"    "t"          "P.Value"    "adj.P.Val" 
## [11] "B"

colnames(group_1_vs_3)

##  [1] "Assay"      "Gene.ID"    "Uniprot.ID" "logFC"      "CI.L"      
##  [6] "CI.R"       "AveExpr"    "t"          "P.Value"    "adj.P.Val" 
## [11] "B"

colnames(group_2_vs_3)

##  [1] "Assay"      "Gene.ID"    "Uniprot.ID" "logFC"      "CI.L"      
##  [6] "CI.R"       "AveExpr"    "t"          "P.Value"    "adj.P.Val" 
## [11] "B"

colnames(control_vs_case_DE_results)

##  [1] "Assay"      "Gene.ID"    "Uniprot.ID" "logFC"      "CI.L"      
##  [6] "CI.R"       "AveExpr"    "t"          "P.Value"    "adj.P.Val" 
## [11] "B"

colnames(group1_longitudinal_DE_results)

##  [1] "Assay"      "Gene.ID"    "Uniprot.ID" "logFC"      "CI.L"      
##  [6] "CI.R"       "AveExpr"    "t"          "P.Value"    "adj.P.Val" 
## [11] "B"

colnames(group3_longitudinal_DE_results)

##  [1] "Assay"      "Gene.ID"    "Uniprot.ID" "logFC"      "CI.L"      
##  [6] "CI.R"       "AveExpr"    "t"          "P.Value"    "adj.P.Val" 
## [11] "B"

# add comparison
group_0_vs_1$Comparison<-"Control_v_Mild"
group_0_vs_2$Comparison<-"Control_v_Severe"
group_0_vs_3$Comparison<-"Control_v_Critical"
group_1_vs_2$Comparison<-"Mild_v_Severe"
group_1_vs_3$Comparison<-"Mild_v_Critical"
group_2_vs_3$Comparison<-"Severe_v_Critical"
control_vs_case_DE_results$Comparison<-"Control_v_Case"
group1_longitudinal_DE_results$Comparison<-"Longitudinal_in_Mild"
group3_longitudinal_DE_results$Comparison<-"Longitudinal_in_Critical"

# move rownames to column
# add comparison
group_0_vs_1$OLINK.ID<-rownames(group_0_vs_1)
group_0_vs_2$OLINK.ID<-rownames(group_0_vs_2)
group_0_vs_3$OLINK.ID<-rownames(group_0_vs_3)
group_1_vs_2$OLINK.ID<-rownames(group_1_vs_2)
group_1_vs_3$OLINK.ID<-rownames(group_1_vs_3)
group_2_vs_3$OLINK.ID<-rownames(group_2_vs_3)
control_vs_case_DE_results$OLINK.ID<-rownames(control_vs_case_DE_results)
group1_longitudinal_DE_results$OLINK.ID<-rownames(group1_longitudinal_DE_results)
group3_longitudinal_DE_results$OLINK.ID<-rownames(group3_longitudinal_DE_results)

# merge
Full_results<-rbind(group_0_vs_1,
                    group_0_vs_2,
                    group_0_vs_3,
                    group_1_vs_2,
                    group_1_vs_3,
                    group_2_vs_3,
                    control_vs_case_DE_results,
                    group1_longitudinal_DE_results,
                    group3_longitudinal_DE_results)


# Adding correlation within groups
# Full_results$cor.with.days<-ifelse(Full_results$OLINK.ID %in% grp_cor_merged$OLINK.ID, "Yes", "No")

# rearrange columns
Full_results<-Full_results[c(12, 1, 13, 2, 3, 4, 5, 6, 7, 9, 10)]

dim(Full_results)

## [1] 3195   11

head(Full_results)

##              Comparison          Assay OLINK.ID        Gene.ID Uniprot.ID
## OID00981 Control_v_Mild         Immune OID00981            NF2     P35240
## OID00374 Control_v_Mild      Neurology OID00374           MANF     P55145
## OID00979 Control_v_Mild         Immune OID00979          BIRC2     Q13490
## OID00401 Control_v_Mild Cardiovascular OID00401 PDGF subunit B     P01127
## OID01011 Control_v_Mild         Immune OID01011          DAPP1     Q9UN19
## OID00936 Control_v_Mild         Immune OID00936        PPP1R9B     Q96SB3
##              logFC      CI.L      CI.R   AveExpr      P.Value    adj.P.Val
## OID00981 -4.621684 -4.802838 -4.440531  1.768762 7.250209e-45 2.573824e-42
## OID00374 -2.854449 -3.076361 -2.632537  7.973475 1.427848e-30 2.534430e-28
## OID00979 -1.735473 -1.901240 -1.569707  1.334477 1.786336e-26 2.113831e-24
## OID00401 -3.228965 -3.559097 -2.898833 10.312451 2.001978e-25 1.776755e-23
## OID01011 -6.619309 -7.314962 -5.923657  5.966673 1.260887e-24 8.952300e-23
## OID00936 -5.279689 -5.864442 -4.694936  4.173545 1.254514e-23 6.528425e-22

#save 
saveRDS(Full_results, file=paste(output_dir, "Result/Full_results.RDS", sep="/"))
write.table(Full_results, file=paste(output_dir, "Result/Full_results.txt", sep="/"), quote=F, sep="\t", row.names = F)

Enrichment test

output files for enrichment test. Using uniprot ID

# list of background
write.table(assay_data_clean$Uniprot.ID, file=paste(output_dir, "/Result/For_enrichment_test/background_list.txt", sep="/"), row.names = F, col.names = F, quote=F)

# group_0_vs_1
write.table(subset(group_0_vs_1, group_0_vs_1$adj.P.Val<=0.05)[3], 
            file=paste(output_dir, "/Result/For_enrichment_test/group_0_vs_1.txt", sep="/"), 
            row.names = F, 
            col.names = F, 
            quote=F)

# group_0_vs_2
write.table(subset(group_0_vs_2, group_0_vs_2$adj.P.Val<=0.05)[3], 
            file=paste(output_dir, "/Result/For_enrichment_test/group_0_vs_2.txt", sep="/"), 
            row.names = F, 
            col.names = F, 
            quote=F)


# group_0_vs_3
write.table(subset(group_0_vs_3, group_0_vs_3$adj.P.Val<=0.05)[3], 
            file=paste(output_dir, "/Result/For_enrichment_test/group_0_vs_3.txt", sep="/"), 
            row.names = F, 
            col.names = F, 
            quote=F)

# group_1_vs_2
write.table(subset(group_1_vs_2, group_1_vs_2$adj.P.Val<=0.05)[3], 
            file=paste(output_dir, "/Result/For_enrichment_test/group_1_vs_2.txt", sep="/"), 
            row.names = F, 
            col.names = F, 
            quote=F)

# group_1_vs_3
write.table(subset(group_1_vs_3, group_1_vs_3$adj.P.Val<=0.05)[3], 
            file=paste(output_dir, "/Result/For_enrichment_test/group_1_vs_3.txt", sep="/"), 
            row.names = F, 
            col.names = F, 
            quote=F)

# group_2_vs_3
write.table(subset(group_2_vs_3, group_2_vs_3$adj.P.Val<=0.05)[3], 
            file=paste(output_dir, "/Result/For_enrichment_test/group_2_vs_3.txt", sep="/"), 
            row.names = F, 
            col.names = F, 
            quote=F)

# control_vs_case_DE_results
write.table(subset(control_vs_case_DE_results, control_vs_case_DE_results$adj.P.Val<=0.05)[3], 
            file=paste(output_dir, "/Result/For_enrichment_test/control_vs_case_DE_results.txt", sep="/"), 
            row.names = F, 
            col.names = F, 
            quote=F)

# group1_longitudinal_DE_results
write.table(subset(group1_longitudinal_DE_results, group1_longitudinal_DE_results$adj.P.Val<=0.05)[3], 
            file=paste(output_dir, "/Result/For_enrichment_test/group1_longitudinal_DE_results.txt", sep="/"), 
            row.names = F, 
            col.names = F, 
            quote=F)

# group3_longitudinal_DE_results
write.table(subset(group3_longitudinal_DE_results, group3_longitudinal_DE_results$adj.P.Val<=0.05)[3], 
            file=paste(output_dir, "/Result/For_enrichment_test/group3_longitudinal_DE_results.txt", sep="/"), 
            row.names = F, 
            col.names = F, 
            quote=F)

number of DE across analyses

# function to calculate number of sig proteins

summarise_DE_by_p<-function(ED_results, analysis_name){
  # emtpy data frame - 1 column, 3 rows
  results<-as.data.frame(matrix(ncol=1, nrow=3))
  #add colname
  colnames(results)<-analysis_name
  # add rownames
  rownames(results)<-c("Down", "NotSig", "Up")
  # number of up regulated proteins
  results[1,1]<-nrow(subset(ED_results, ED_results$adj.P.Val<0.05 & logFC>0))
  # number of non sig
  results[2,1]<-nrow(subset(ED_results, ED_results$adj.P.Val>=0.05))
  # number of down sig regulted proteins
  results[3,1]<-nrow(subset(ED_results, ED_results$adj.P.Val<0.05 & logFC<0))
  return(results)
}

# apply function

group_0_vs_1_DE_count <- summarise_DE_by_p(group_0_vs_1, "Control_v_Mild")
group_0_vs_2_DE_count <- summarise_DE_by_p(group_0_vs_2, "Control_v_Moderate")
group_0_vs_3_DE_count <- summarise_DE_by_p(group_0_vs_3 , "Control_v_Severe")
group_1_vs_2_DE_count <- summarise_DE_by_p(group_1_vs_2 , "Mild_v_Moderate")
group_1_vs_3_DE_count <- summarise_DE_by_p(group_1_vs_3 , "Mild_v_Severe")
group_2_vs_3_DE_count <- summarise_DE_by_p(group_2_vs_3 , "Moderate_v_Severe")
control_vs_case_DE_count <- summarise_DE_by_p(control_vs_case_DE_results , "Control_v_Case")
group1_longitudinal_DE_count <- summarise_DE_by_p(group1_longitudinal_DE_results , "Longitudinal_in_Mild")
group3_longitudinal_DE_count <- summarise_DE_by_p(group3_longitudinal_DE_results , "Longitudinal_in_Severe")

# merg results

DE_result_count<-cbind(group_0_vs_1_DE_count,
                       group_0_vs_2_DE_count,
                       group_0_vs_3_DE_count,
                       group_1_vs_2_DE_count,
                       group_1_vs_3_DE_count,
                       group_2_vs_3_DE_count,
                       control_vs_case_DE_count,
                       group1_longitudinal_DE_count,
                       group3_longitudinal_DE_count)

# check
DE_result_count

##        Control_v_Mild Control_v_Moderate Control_v_Severe Mild_v_Moderate
## Down               42                  8               82              97
## NotSig            208                290              168             209
## Up                105                 57              105              49
##        Mild_v_Severe Moderate_v_Severe Control_v_Case Longitudinal_in_Mild
## Down             158                63            120                   10
## NotSig           158               291             86                  342
## Up                39                 1            149                    3
##        Longitudinal_in_Severe
## Down                        3
## NotSig                    349
## Up                          3

Proteins sig across control->mild->moderate->severe

These proteins are sig between:

1. control -> mild
2. control -> moderate
3. control -> severe

# extract proteins sig diff in group 0->1->2->3
proteins_sig_across_all<-subset(Full_results, Full_results$Comparison=="Control_v_Mild" & Full_results$adj.P.Val<=0.05 |
  Full_results$Comparison=="Mild_v_Critical" & Full_results$adj.P.Val<=0.05 |
  Full_results$Comparison=="Severe_v_Critical" & Full_results$adj.P.Val<=0.05)

# keep only those that are repeated 3 times
proteins_sig_across_all<-subset(proteins_sig_across_all, proteins_sig_across_all$OLINK.ID %in% subset(as.data.frame(table(proteins_sig_across_all$OLINK.ID)), Freq==3)$Var1)

# sort by OLINK
proteins_sig_across_all<-proteins_sig_across_all[order(proteins_sig_across_all$OLINK.ID),]

# check
head(proteins_sig_across_all)

##                  Comparison     Assay OLINK.ID Gene.ID Uniprot.ID     logFC
## OID00300     Control_v_Mild Neurology OID00300  SCARB2     Q14108 0.5942784
## OID003004   Mild_v_Critical Neurology OID00300  SCARB2     Q14108 0.6278560
## OID003005 Severe_v_Critical Neurology OID00300  SCARB2     Q14108 0.7269831
## OID00315     Control_v_Mild Neurology OID00315 SIGLEC1     Q9BZZ2 1.7679780
## OID003154   Mild_v_Critical Neurology OID00315 SIGLEC1     Q9BZZ2 0.4064055
## OID003155 Severe_v_Critical Neurology OID00315 SIGLEC1     Q9BZZ2 0.4096786
##                 CI.L      CI.R  AveExpr      P.Value    adj.P.Val
## OID00300  0.23940756 0.9491492 4.593904 1.483496e-03 4.702154e-03
## OID003004 0.18779937 1.0679127 5.259929 6.167083e-03 1.371865e-02
## OID003005 0.33069805 1.1232681 5.682310 7.712754e-04 1.557798e-02
## OID00315  1.22773304 2.3082230 6.017946 2.799808e-08 1.774878e-07
## OID003154 0.06085954 0.7519514 6.787408 2.219098e-02 4.124502e-02
## OID003155 0.13311006 0.6862472 7.220382 5.081584e-03 3.340671e-02

nrow(proteins_sig_across_all)

## [1] 78

# empty list to remove
proteins_sig_across_all_to_remove<-vector()

# keep proteins where logFC is same direction in all analysis
for (x in seq(1, nrow(proteins_sig_across_all),3)) {
  # extract fold change by 3
  numbers<-proteins_sig_across_all$logFC[x:(x+2)]
  if(all(numbers>0)==F & all(numbers<0)==F) {
    proteins_sig_across_all_to_remove<-c(proteins_sig_across_all_to_remove, proteins_sig_across_all$OLINK.ID[x])
  }
}

# remove
proteins_sig_across_all<-subset(proteins_sig_across_all, !(proteins_sig_across_all$OLINK.ID %in% proteins_sig_across_all_to_remove))
nrow(proteins_sig_across_all)  

## [1] 57

# sort b =p
proteins_sig_across_all<-proteins_sig_across_all[order(proteins_sig_across_all$adj.P.Val),]
proteins_sig_across_all

##                  Comparison          Assay OLINK.ID Gene.ID Uniprot.ID
## OID009474   Mild_v_Critical         Immune OID00947     IL6     P05231
## OID004824   Mild_v_Critical   Inflammation OID00482     IL6     P05231
## OID009854   Mild_v_Critical         Immune OID00985   CKAP4     Q07065
## OID003904   Mild_v_Critical Cardiovascular OID00390     IL6     P05231
## OID004064   Mild_v_Critical Cardiovascular OID00406   Gal-9     O00182
## OID003894   Mild_v_Critical Cardiovascular OID00389  IL-1ra     P18510
## OID009654   Mild_v_Critical         Immune OID00965  LILRB4     Q8NHJ6
## OID00315     Control_v_Mild      Neurology OID00315 SIGLEC1     Q9BZZ2
## OID004644   Mild_v_Critical Cardiovascular OID00464    CA5A     P35218
## OID005184   Mild_v_Critical   Inflammation OID00518   PD-L1     Q9NZQ7
## OID005174   Mild_v_Critical   Inflammation OID00517 IL-18R1     Q13478
## OID00441     Control_v_Mild Cardiovascular OID00441    MMP7     P09237
## OID005354   Mild_v_Critical   Inflammation OID00535  CXCL10     P02778
## OID01015     Control_v_Mild         Immune OID01015   LAMP3     Q9UQV4
## OID00406     Control_v_Mild Cardiovascular OID00406   Gal-9     O00182
## OID00535     Control_v_Mild   Inflammation OID00535  CXCL10     P02778
## OID00464     Control_v_Mild Cardiovascular OID00464    CA5A     P35218
## OID00389     Control_v_Mild Cardiovascular OID00389  IL-1ra     P18510
## OID004844   Mild_v_Critical   Inflammation OID00484   MCP-1     P13500
## OID003834   Mild_v_Critical Cardiovascular OID00383  SLAMF7     Q9NQ25
## OID005175 Severe_v_Critical   Inflammation OID00517 IL-18R1     Q13478
## OID003895 Severe_v_Critical Cardiovascular OID00389  IL-1ra     P18510
## OID00484     Control_v_Mild   Inflammation OID00484   MCP-1     P13500
## OID00552     Control_v_Mild   Inflammation OID00552  CX3CL1     P78423
## OID005524   Mild_v_Critical   Inflammation OID00552  CX3CL1     P78423
## OID00300     Control_v_Mild      Neurology OID00300  SCARB2     Q14108
## OID00517     Control_v_Mild   Inflammation OID00517 IL-18R1     Q13478
## OID00985     Control_v_Mild         Immune OID00985   CKAP4     Q07065
## OID004415 Severe_v_Critical Cardiovascular OID00441    MMP7     P09237
## OID010154   Mild_v_Critical         Immune OID01015   LAMP3     Q9UQV4
## OID00482     Control_v_Mild   Inflammation OID00482     IL6     P05231
## OID00965     Control_v_Mild         Immune OID00965  LILRB4     Q8NHJ6
## OID005185 Severe_v_Critical   Inflammation OID00518   PD-L1     Q9NZQ7
## OID00390     Control_v_Mild Cardiovascular OID00390     IL6     P05231
## OID004254   Mild_v_Critical Cardiovascular OID00425   MERTK     Q12866
## OID005525 Severe_v_Critical   Inflammation OID00552  CX3CL1     P78423
## OID003004   Mild_v_Critical      Neurology OID00300  SCARB2     Q14108
## OID003005 Severe_v_Critical      Neurology OID00300  SCARB2     Q14108
## OID004645 Severe_v_Critical Cardiovascular OID00464    CA5A     P35218
## OID005355 Severe_v_Critical   Inflammation OID00535  CXCL10     P02778
## OID003905 Severe_v_Critical Cardiovascular OID00390     IL6     P05231
## OID010155 Severe_v_Critical         Immune OID01015   LAMP3     Q9UQV4
## OID009655 Severe_v_Critical         Immune OID00965  LILRB4     Q8NHJ6
## OID009855 Severe_v_Critical         Immune OID00985   CKAP4     Q07065
## OID004065 Severe_v_Critical Cardiovascular OID00406   Gal-9     O00182
## OID004825 Severe_v_Critical   Inflammation OID00482     IL6     P05231
## OID00518     Control_v_Mild   Inflammation OID00518   PD-L1     Q9NZQ7
## OID003835 Severe_v_Critical Cardiovascular OID00383  SLAMF7     Q9NQ25
## OID003155 Severe_v_Critical      Neurology OID00315 SIGLEC1     Q9BZZ2
## OID00383     Control_v_Mild Cardiovascular OID00383  SLAMF7     Q9NQ25
## OID00947     Control_v_Mild         Immune OID00947     IL6     P05231
## OID00425     Control_v_Mild Cardiovascular OID00425   MERTK     Q12866
## OID003154   Mild_v_Critical      Neurology OID00315 SIGLEC1     Q9BZZ2
## OID009475 Severe_v_Critical         Immune OID00947     IL6     P05231
## OID004414   Mild_v_Critical Cardiovascular OID00441    MMP7     P09237
## OID004255 Severe_v_Critical Cardiovascular OID00425   MERTK     Q12866
## OID004845 Severe_v_Critical   Inflammation OID00484   MCP-1     P13500
##               logFC       CI.L      CI.R   AveExpr      P.Value    adj.P.Val
## OID009474 4.4228712 3.62937566 5.2163667  4.599221 1.140249e-14 1.092434e-12
## OID004824 4.4630021 3.64214137 5.2838627  4.752515 1.995853e-14 1.417056e-12
## OID009854 1.5962310 1.27018936 1.9222726  5.770290 7.610724e-13 4.297903e-11
## OID003904 4.4796139 3.50733004 5.4518977  5.603931 4.026385e-12 1.786708e-10
## OID004064 0.8021345 0.60998283 0.9942861  8.703112 7.297999e-11 2.294098e-09
## OID003894 1.8647427 1.41170028 2.3177852  6.414466 1.085524e-10 2.964316e-09
## OID009654 1.7068637 1.21500199 2.1987254  4.540222 1.030364e-08 1.524080e-07
## OID00315  1.7679780 1.22773304 2.3082230  6.017946 2.799808e-08 1.774878e-07
## OID004644 1.7929475 1.24844215 2.3374529  3.698926 3.228637e-08 3.697310e-07
## OID005184 1.0393479 0.71253313 1.3661627  6.175398 7.088193e-08 6.989746e-07
## OID005174 0.8407626 0.57270612 1.1088191  9.041958 9.619410e-08 9.201371e-07
## OID00441  1.5051666 0.96348315 2.0468501  8.902822 9.320688e-07 4.726921e-06
## OID005354 1.8266501 1.18902831 2.4642720 11.598658 6.382902e-07 5.035400e-06
## OID01015  1.7709154 1.08382405 2.4580068  4.448952 4.028238e-06 1.958938e-05
## OID00406  0.9575807 0.53472696 1.3804345  8.066001 3.388933e-05 1.503839e-04
## OID00535  2.4225698 1.33514026 3.5099994 10.083450 4.344471e-05 1.904058e-04
## OID00464  1.5867562 0.87292894 2.3005834  2.506874 4.490463e-05 1.944042e-04
## OID00389  1.4497846 0.77962760 2.1199416  5.217257 6.690251e-05 2.729930e-04
## OID004844 1.4299050 0.68662672 2.1731832 12.279563 3.377112e-04 1.120444e-03
## OID003834 1.0142716 0.48604464 1.5424986  3.951016 3.454442e-04 1.135488e-03
## OID005175 0.5789307 0.35997899 0.7978824  9.615739 7.829294e-06 1.389700e-03
## OID003895 1.0461594 0.62673486 1.4655839  7.317076 1.846293e-05 2.184780e-03
## OID00484  1.0033769 0.44533591 1.5614179 11.308540 6.988529e-04 2.444943e-03
## OID00552  0.6248476 0.26378658 0.9859087  4.096594 1.054237e-03 3.436315e-03
## OID005524 0.4964143 0.19675514 0.7960735  4.487630 1.692105e-03 4.585476e-03
## OID00300  0.5942784 0.23940756 0.9491492  4.593904 1.483496e-03 4.702154e-03
## OID00517  0.7590735 0.30267341 1.2154735  8.385347 1.577485e-03 4.955816e-03
## OID00985  0.5208326 0.20643949 0.8352257  4.652782 1.649907e-03 5.137868e-03
## OID004415 0.7183900 0.39560657 1.0411734  9.855883 8.578543e-05 6.090766e-03
## OID010154 0.8559165 0.32147663 1.3903563  5.362570 2.340196e-03 6.108601e-03
## OID00482  1.0826512 0.41449025 1.7508122  2.374035 2.029106e-03 6.236336e-03
## OID00965  0.8842353 0.33860331 1.4298673  3.297511 2.037789e-03 6.236336e-03
## OID005185 0.4927042 0.26255432 0.7228542  6.818685 1.389122e-04 7.044832e-03
## OID00390  1.1779535 0.44011449 1.9157924  3.145305 2.331035e-03 7.072797e-03
## OID004254 0.3441327 0.10710131 0.5811640  6.805271 5.364707e-03 1.244752e-02
## OID005525 0.6942100 0.34031318 1.0481069  4.788656 3.810444e-04 1.352708e-02
## OID003004 0.6278560 0.18779937 1.0679127  5.259929 6.167083e-03 1.371865e-02
## OID003005 0.7269831 0.33069805 1.1232681  5.682310 7.712754e-04 1.557798e-02
## OID004645 1.1593538 0.53270509 1.7860025  4.600708 7.078011e-04 1.557798e-02
## OID005355 0.6233459 0.26836309 0.9783287 13.038169 1.185146e-03 1.618180e-02
## OID003905 1.7990829 0.76797313 2.8301927  7.816368 1.259427e-03 1.655913e-02
## OID010155 0.6611489 0.27022973 1.0520681  5.922254 1.682039e-03 1.926206e-02
## OID009655 0.7328342 0.29089556 1.1747728  5.391841 2.008154e-03 2.160287e-02
## OID009855 0.8178668 0.30931307 1.3264206  6.649743 2.620216e-03 2.383006e-02
## OID004065 0.2754089 0.09731255 0.4535052  9.245245 3.626564e-03 3.065310e-02
## OID004825 1.7094165 0.59904232 2.8197907  6.926543 3.759657e-03 3.103903e-02
## OID00518  0.5913870 0.13486657 1.0479075  5.495612 1.214420e-02 3.124052e-02
## OID003835 0.8217724 0.27455457 1.3689902  4.415489 4.573717e-03 3.334589e-02
## OID003155 0.4096786 0.13311006 0.6862472  7.220382 5.081584e-03 3.340671e-02
## OID00383  0.8836114 0.18979563 1.5774272  3.188798 1.358490e-02 3.444742e-02
## OID00947  0.7428272 0.15726051 1.3283939  2.251074 1.395526e-02 3.488815e-02
## OID00425  0.4396680 0.08712706 0.7922090  6.391581 1.553126e-02 3.828886e-02
## OID003154 0.4064055 0.06085954 0.7519514  6.787408 2.219098e-02 4.124502e-02
## OID009475 1.7433857 0.51963356 2.9671379  6.629754 6.784581e-03 4.379139e-02
## OID004414 0.4945356 0.06584914 0.9232220  9.762914 2.470363e-02 4.513650e-02
## OID004255 0.3581592 0.09879963 0.6175187  7.065892 8.451639e-03 4.839245e-02
## OID004845 1.1567006 0.32369174 1.9897094 12.988422 8.134528e-03 4.839245e-02

# multiple gene names
#proteins_sig_across_all$Gene.ID<-droplevels(proteins_sig_across_all$Gene.ID)
table(proteins_sig_across_all$Gene.ID)

## 
##               4E-BP1                 ACE2                  ADA 
##                    0                    0                    0 
##            ADAM-TS13              ADAM 22              ADAM 23 
##                    0                    0                    0 
##                  ADM                 AGRP         Alpha-2-MRAP 
##                    0                    0                    0 
##                 AMBP               ANGPT1                 AREG 
##                    0                    0                    0 
##                 ARNT                 ARTN                AXIN1 
##                    0                    0                    0 
##                BACH1                 BCAN             Beta-NGF 
##                    0                    0                    0 
##                BIRC2                BMP-4                BMP-6 
##                    0                    0                    0 
##                  BNP                  BOC               BTN3A2 
##                    0                    0                    0 
##                 CA5A                CADM3               CASP-8 
##                    3                    0                    0 
##                CCL11                CCL17                CCL19 
##                    0                    0                    0 
##                CCL20                CCL23                CCL25 
##                    0                    0                    0 
##                CCL28                 CCL3                 CCL4 
##                    0                    0                    0 
##                CD200              CD200R1                CD244 
##                    0                    0                    0 
##                 CD28                 CD38                  CD4 
##                    0                    0                    0 
##                 CD40               CD40-L                  CD5 
##                    0                    0                    0 
##                  CD6                 CD83                 CD84 
##                    0                    0                    0 
##                 CD8A                CDCP1                 CDH3 
##                    0                    0                    0 
##                 CDH6                 CDSN              CEACAM8 
##                    0                    0                    0 
##                CKAP4              CLEC10A               CLEC1B 
##                    3                    0                    0 
##               CLEC4A               CLEC4C               CLEC4D 
##                    0                    0                    0 
##               CLEC4G               CLEC6A               CLEC7A 
##                    0                    0                    0 
##                CLM-1                CLM-6                CNTN5 
##                    0                    0                    0 
##              CNTNAP2                 CPA2                  CPM 
##                    0                    0                    0 
##                CRTAM                CSF-1                 CST5 
##                    0                    0                    0 
##                 CTRC                 CTSC                CTSL1 
##                    0                    0                    0 
##                 CTSS               CX3CL1                CXADR 
##                    0                    3                    0 
##                CXCL1               CXCL10               CXCL11 
##                    0                    3                    0 
##               CXCL12                CXCL5                CXCL6 
##                    0                    0                    0 
##                CXCL9                DAPP1               DCBLD2 
##                    0                    0                    0 
##                  DCN                DCTN1                 DDR1 
##                    0                    0                    0 
##                DDX58                DECR1                 DFFA 
##                    0                    0                    0 
##                 DGKZ                Dkk-1                Dkk-4 
##                    0                    0                    0 
##                 DNER                DPP10               DRAXIN 
##                    0                    0                    0 
##                EDA2R                 EDAR                EFNA4 
##                    0                    0                    0 
##                EGLN1               EIF4G1                EIF5A 
##                    0                    0                    0 
##              EN-RAGE                EPHB6                  EZR 
##                    0                    0                    0 
##                FABP2                FAM3B                FcRL2 
##                    0                    0                    0 
##                FCRL3                FCRL6               FGF-19 
##                    0                    0                    0 
##               FGF-21               FGF-23                FGF-5 
##                    0                    0                    0 
##                 FGF2                FLRT2                Flt3L 
##                    0                    0                    0 
##                   FS                FXYD5                G-CSF 
##                    0                    0                    0 
##                gal-8                Gal-9               GALNT3 
##                    0                    3                    0 
##                 GCP5                GDF-2                GDF-8 
##                    0                    0                    0 
##                 GDNF        GDNFR-alpha-3          GFR-alpha-1 
##                    0                    0                    0 
##                   GH                  GIF                 GLB1 
##                    0                    0                    0 
##                 GLO1       GM-CSF-R-alpha                   GT 
##                    0                    0                    0 
##                 GZMA                 HAGH                HAOX1 
##                    0                    0                    0 
##               HB-EGF                HCLS1               HEXIM1 
##                    0                    0                    0 
##                  HGF                 HNMT                 HO-1 
##                    0                    0                    0 
##               hOSCAR              HSD11B1               HSP 27 
##                    0                    0                    0 
##                 ICA1                 IDUA            IFN-gamma 
##                    0                    0                    0 
##               IFNLR1 IgG Fc receptor II-b           IL-1 alpha 
##                    0                    0                    0 
##              IL-10RA              IL-10RB               IL-12B 
##                    0                    0                    0 
##              IL-15RA               IL-17A               IL-17C 
##                    0                    0                    0 
##               IL-17D              IL-18R1               IL-1ra 
##                    0                    3                    3 
##                IL-20              IL-20RA            IL-22 RA1 
##                    0                    0                    0 
##                IL-24                IL-27               IL-2RB 
##                    0                    0                    0 
##               IL-4RA          IL-5R-alpha                 IL10 
##                    0                    0                    0 
##                 IL12              IL12RB1                 IL13 
##                    0                    0                    0 
##                 IL16                 IL18               IL1RL2 
##                    0                    0                    0 
##                  IL2                 IL33                  IL4 
##                    0                    0                    0 
##                  IL5                  IL6                  IL7 
##                    0                    9                    0 
##                  IL8                IRAK1                IRAK4 
##                    0                    0                    0 
##                 IRF9               ITGA11                ITGA6 
##                    0                    0                    0 
##             ITGB1BP2                ITGB6                ITM2A 
##                    0                    0                    0 
##                JAM-B                  JUN                 KIM1 
##                    0                    0                    0 
##                KLRD1                KPNA1                KRT19 
##                    0                    0                    0 
##                 KYNU                 LAG3               LAIR-2 
##                    0                    0                    0 
##                LAMP3       LAP TGF-beta-1                  LAT 
##                    3                    0                    0 
##                 LAYN                  LEP                  LIF 
##                    0                    0                    0 
##                LIF-R               LILRB4                LOX-1 
##                    0                    3                    0 
##                  LPL                  LXN                 LY75 
##                    0                    0                    0 
##                 MANF                 MAPT                MARCO 
##                    0                    0                    0 
##                MASP1                MATN3                MCP-1 
##                    0                    0                    3 
##                MCP-2                MCP-3                MCP-4 
##                    0                    0                    0 
##                MDGA1                MERTK                 MGMT 
##                    0                    3                    0 
##                MILR1                MMP-1               MMP-10 
##                    0                    0                    0 
##                MMP12                 MMP7                 MSR1 
##                    0                    3                    0 
##              N-CDase               N2DL-2                 NAAA 
##                    0                    0                    0 
##                 NBL1                 NCAN                 NCR1 
##                    0                    0                    0 
##                 NEMO                  NEP                  NF2 
##                    0                    0                    0 
##               NFATC3               NMNAT1               Nr-CAM 
##                    0                    0                    0 
##                 NRP2                 NRTN                 NT-3 
##                    0                    0                    0 
##                 NTF4                NTRK2                NTRK3 
##                    0                    0                    0 
##                  OPG                  OSM                PADI2 
##                    0                    0                    0 
##                PAPPA                PAR-1               PARP-1 
##                    0                    0                    0 
##                PD-L1                PD-L2         PDGF-R-alpha 
##                    3                    0                    0 
##       PDGF subunit B                  PGF                 PIgR 
##                    0                    0                    0 
##              PIK3AP1               PLXNA4               PLXNB1 
##                    0                    0                    0 
##               PLXNB3              PPP1R9B                PRDX1 
##                    0                    0                    0 
##                PRDX3                PRDX5                PRELP 
##                    0                    0                    0 
##                PRKCQ               PRSS27                PRSS8 
##                    0                    0                    0 
##                 PRTG               PSGL-1                PSIP1 
##                    0                    0                    0 
##                PTH1R                 PTX3                  PVR 
##                    0                    0                    0 
##                 RAGE                  REN                 RGMA 
##                    0                    0                    0 
##                 RGMB                ROBO2                RSPO1 
##                    0                    0                    0 
##               SCARA5               SCARB2               SCARF2 
##                    0                    3                    0 
##                  SCF            SERPINA12               sFRP-3 
##                    0                    0                    0 
##                SH2B3               SH2D1A             Siglec-9 
##                    0                    0                    0 
##              SIGLEC1                SIRT2                 SIT1 
##                    3                    0                    0 
##                 SKR3               SLAMF1               SLAMF7 
##                    0                    0                    3 
##                SMOC2                SMPD1                 SOD2 
##                    0                    0                    0 
##                SORT1               SPOCK1                SPON2 
##                    0                    0                    0 
##                SPRY2                  SRC                SRPK2 
##                    0                    0                    0 
##                ST1A1               STAMBP                 STC1 
##                    0                    0                    0 
##                 STK4                 TANK                   TF 
##                    0                    0                    0 
##            TGF-alpha                 TGM2                THBS2 
##                    0                    0                    0 
##                 THPO                THY 1                 TIE2 
##                    0                    0                    0 
##                   TM              TMPRSS5                 TN-R 
##                    0                    0                    0 
##                  TNF                 TNFB            TNFRSF10A 
##                    0                    0                    0 
##            TNFRSF11A            TNFRSF12A            TNFRSF13B 
##                    0                    0                    0 
##             TNFRSF21              TNFRSF9              TNFSF14 
##                    0                    0                    0 
##               TPSAB1                TRAF2                TRAIL 
##                    0                    0                    0 
##             TRAIL-R2               TRANCE                TREM1 
##                    0                    0                    0 
##               TRIM21                TRIM5                 TSLP 
##                    0                    0                    0 
##                TWEAK                UNC5C                  uPA 
##                    0                    0                    0 
##                VEGFA                VEGFD                VSIG2 
##                    0                    0                    0 
##                 VWC2              WFIKKN1                 XCL1 
##                    0                    0                    0 
##               ZBTB16 
##                    0

plot

# plot most anti-correlated
ggplot(data=protein_data_clean[c(1,(grep("OID00390", colnames(protein_data_clean))))], 
         aes(x=as.character(Group), y=OID00390, fill=as.character(Group))) +
    geom_boxplot()

ggplot(data=protein_data_clean[c(1,(grep("OID00985", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00985, fill=as.character(Group))) +
  geom_boxplot()

## Warning: Removed 1 rows containing non-finite values (stat_boxplot).

ggplot(data=protein_data_clean[c(1,(grep("OID00406", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00406, fill=as.character(Group))) +
  geom_boxplot()

ggplot(data=protein_data_clean[c(1,(grep("OID00389", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00389, fill=as.character(Group))) +
  geom_boxplot()

ggplot(data=protein_data_clean[c(1,(grep("OID00965", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00965, fill=as.character(Group))) +
  geom_boxplot()

## Warning: Removed 1 rows containing non-finite values (stat_boxplot).

ggplot(data=protein_data_clean[c(1,(grep("OID00518", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00518, fill=as.character(Group))) +
  geom_boxplot()

Summary:

• 9 proteins sig dif
• 6 unique:
  • CKAP4
  • Gal-9
  • IL-1ra
  • IL6
  • LILRB4
  • PD-L1

Manuscript plots

plots

tiff(file=paste(paste(output_dir, "plots/NF2.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00981", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00981, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("NF2 Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()

NF2 <- protein_data_clean[c(1,2,3,4,5,6, (grep("OID00981", colnames(protein_data_clean))))]

# order
NF2<-NF2[order(NF2$Group),]

# add level
NF2$sample<-1:nrow(NF2)

tiff(file=paste(paste(output_dir, "plots/NF2_individual.tiff", sep="/")), width = 10, height = 10, units = 'in', res = 300)
ggplot(data= NF2, aes(x=sample, y=OID00981, fill=as.character(Group))) +
  geom_bar(stat="identity") +
  labs(x = "Samples",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("NF2 Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()

tiff(file=paste(paste(output_dir, "plots/CKAP4.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00985", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00985, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("CKAP4 Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()

tiff(file=paste(paste(output_dir, "plots/Gal-9.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00406", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00406, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("Gal-9 Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()

tiff(file=paste(paste(output_dir, "plots/IL-1ra.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00389", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00389, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("IL-1ra Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()

tiff(file=paste(paste(output_dir, "plots/PD-L1.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00518", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00518, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("PD-L1 Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()

tiff(file=paste(paste(output_dir, "plots/LILRB4.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00965", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00965, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("LILRB4 Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()


tiff(file=paste(paste(output_dir, "plots/EDA2R.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00370", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00370, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("EDA2R Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()


tiff(file=paste(paste(output_dir, "plots/SCARB2.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00300", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00300, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("SCARB2 Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()

tiff(file=paste(paste(output_dir, "plots/MANF.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00374", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00374, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("MANF Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()

tiff(file=paste(paste(output_dir, "plots/LAT.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00371", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00371, fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("LAT Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()

tiff(file=paste(paste(output_dir, "plots/SIGLEC1.tiff", sep="/")))
ggplot(data=protein_data_clean[c(1,(grep("OID00315", colnames(protein_data_clean))))], 
       aes(x=as.character(Group), y=OID00315 , fill=as.character(Group))) +
  geom_boxplot() +
  labs(x = "Disease Group",
       y = "Expression (NPX)") + 
  scale_fill_discrete(name = "Disease Group", labels = c("Control", "Mild", "Severe", "Critical")) +
  scale_x_discrete(labels=c("0" = "Control", "1" = "Mild", "2" = "Severe", "3" = "Critical")) +
  ggtitle("SIGLEC1 Expression")+
  theme(plot.title = element_text(hjust = 0.5))
dev.off()