isoMIGA: Expression and Splicing QTL Summary Statistics in Human Microglia

https://github.com/RajLabMSSM/isoMiGA 

Full associations and top association summary statistics for QTLs mapped in a multi-ethnic meta-analysis of human microglia, as part of the isoform-centric Microglia Genomic Atlas (isoMiGA) project.

Sample size = 555 samples from 391 unique donors

6 cohorts (Raj MFG, Raj STG, Raj, SVZ, Raj, THA, Roussos, Gaffney) meta-analysed using the linear mixed model random-effects meta-analysis software mmQTL (PMID: 35058635).

Each file has the following naming convention:

{REFERENCE}_{PHENOTYPE}_{ASSOC}.tsv.gz

References

The following two transcript references were used for mapping phenotypes:

1. GENCODE - GENCODE v38 comprehensive transcripts (https://www.gencodegenes.org/human/release_38.html)

2. Union - a union of all GENCODE v38 transcripts with an additional 35,879 novel transcripts identified from long-read RNA-seq in 30 human microglia samples. GTF to be uploaded in Zenodo.

Phenotypes

The following phenotypes were tested for genetic association:

• expression: total gene expression of each gene following voom normalization of read counts.
• transcript: transcript usage - each transcript expression normalized by TPM divided by the total TPM for that gene
• leafcutter: junction usage using the Leafcutter framework to count and cluster intron-splicing junction reads. Each phenotype is the relative usage of the intron against the total count of the introns within that cluster.
• SUPPA_A3: alternate 3' splice site usage, as identified by SUPPA2 from transcript TPMs
• SUPPA_A5: alternate 5' splice site usage
• SUPPA_AF: alternate first exon usage
• SUPPA_AL: alternate last exon usage
• SUPPA_RI: intron retention usage
• SUPPA_SE: exon skipping usage

All phenotype matrices were scaled and centred and then quantile normalized.

Associations

Top associations (top_assoc.tsv.gz) list the SNP-feature pair with the lowest adjusted P-value (qval) for that feature.

Full associations (full_assoc.tsv.gz) list all tested SNP-feature pairs.

Data dictionary

The columns of the two association files only differ by the presence of the qval column in the top associations.

feature: the phenotype being tested

variant_id: the genetic variant being tested

chr: chromosome

pos: position (hg38)

ref: reference allele

alt: alternate allele

Allele: the effect allele that the beta is relative to

beta_tissue_0: Gaffney cohort beta

sd_tissue_0: Gaffney cohort standard error

z_tissue_0: Gafney cohort Z-score

beta_tissue_1: Roussos cohort beta

sd_tissue_1: Roussos cohort standard error

z_tissue_1: Roussos cohort Z-core

beta_tissue_2: Raj MFG cohort beta

sd_tissue_2: Raj MFG cohort standard error

z_tissue_2: Raj MFG cohort Z-score

beta_tissue_3: Raj STG cohort beta

sd_tissue_3: Raj STG cohort standard error

z_tissue_3: Raj STG cohort Z-score

beta_tissue_4: Raj SVZ cohort beta

sd_tissue_4: Raj SVZ cohort standard error

z_tissue_4: Raj SVZ cohort Z-score

beta_tissue_5: Raj THA cohort beta

sd_tissue_5: Raj THA cohort standard error

z_tissue_5: Raj THA cohort Z-score

fixed_beta: Fixed effect meta-analysis estimate of the beta

fixed_sd: Fixed effect meta-analysis standard error of the beta

fixed_z: Fixed effect meta-analysis Z-score

Random_Z: Random effect meta-analysis Z-score

Fixed_P: Fixed effect meta-analysis P-value

Random_P: Random effect meta-analysis P-value

Fixed_bonf:  Fixed effect meta-analysis P-value adjusted for the number of variants tested in that feature (Bonferroni)

Random_bonf: Random effect meta-analysis P-value adjusted for the number of variants tested in that feature (Bonferroni)

Fixed_FDR: Fixed effect meta-analysis P-value adjusted for the number of variants tested in that feature (FDR)

Random_FDR: Random effect meta-analysis P-value adjusted for the number of variants tested in that feature (FDR)

qval: Random effect meta-analysis P-value adjusted for the number of variants tested in that feature (FDR) and for the number of features tested in the dataset (Storey's q value)