Atlas of virus specific CD8+ Tcells

Code and Data accompanying the manuscript " "

TS logistic regression model:

• TargetScape_Model_FCS_Input.zip contains the TS data in pickled format to be imported in python
• targetscape-model-paper-20221001.html contains the python code to perform data analysis, model training, testing and interpretation

TAP logistic regression model:

• Generate_Pseudobulk_Markers_ADT_TAP_Cohort.R to generate candidate features by pseudo-bulking surface markers across donors.
• TAP_ParameterBenchmarking_Model_Training_Evaluation.R to perform model training and parameter benchmark.
• TAP_DiscoveryCohort_Epitopes.rds file containing a Seurat object of the TAP-Cohort.
• Cohort_Pseudobulk_ADT_Virus_markers.RDS RDS file containing the candidate markers and their fold-changes/p-values, generated by Generate_Pseudobulk_Markers_ADT_TAP_Cohort.R
• Leave-one-epitope-out-TAP.R: R script to train a model using one epitope as a validation set.

Figures related to Machine Learning approaches:

• Model_Analysis_Figure_Table_Generation.R script to generate Figures 3 and Figure 4 of the main manuscript as well as corresponding Supplementary Figures
• TargetScapeModel_Feature_Importance.csv contains the regression coefficients of the model learned on TargetScape (TS) data.
• TargetScape_Cohort_Logicle.csv contains the actual data of the TS cohort after logicle transformation.
• TargetScape_Validation_Cohort_Logicle.csv contains the actual data of the TS validation cohort after logicle transformation
• TargetScape_Model_Predictions.csv contains predictions of the TS model made on the TS validation cohort
• Performance_Benchmarking_TAP.zip contains all parameter benchmarking performance files generated by the TAP model above while executing TAP_ParameterBenchmarking_Model_Training_Evaluation.R.
• Performance_Selected_Model.txt contains model performance information on the selected TAP model
• TAP_Model_MultinomialElaNetLogRegModel_minCellsPerPatient_30_minCellsPerVirus_200_ResourcePaper.RDS contains the best model trained on TAP data.
• TAP-Validation.RDS contains the Seurat object of the TAP validation cohort.