PathWeb displays results for associations between over 10,000 pathways (gene-sets) and phenotypes derived from ICD billing codes of White British participants of the UK Biobank. GWAS summary-statistics obtained using SAIGE for 1,403 binary phenotypes derived from ICD billing codes have been used in the analysis here. The single-variant summary statistics are available on PheWeb. The association analysis between a pathway and the summary statistics of a phenotype was done using GAUSS which produces a p-value for the association between the phenotype and the pathway (gene-set) and additionally extracts a 'core subset' (CS) genes within the gene-set which drives the association.
We used two categories of pathways in our analysis from Molecular Signatures database (MSigDB v6.2):
Curated pathways (C2) including KEGG, BioCarta (BIOCARTA), REACTOME and chemical genetic perturbations (CGP) and other canonical pathways (OTHER_CANONICAL)
GO pathways (C5) including Biological processes (BIOLOGICAL_PROC), molecular processes (MOLECULAR), cellular processes (CELLULAR).
In the current version, we have displayed resampling p-values for GAUSS using a million resampling iterations. Any p-value < 10-6 is collapsed to a p-value of 10-6. We are working on making the p-values using generalized pareto distribution (GPD) for all phenotypes and gene-sets available. Please stay tuned !!
If you use the results from the analysis please cite the manuscript preprint on bioRXiv
GAUSS R-package and instruction manual is available on GitHub. We will also make the results displayed here freely available (download-able) in the near future.
This site was built with modifictions from the PheWeb package