GCDPipe: risk gene, cell type, and drug ranking for complex traits

Abstract

We introduce a user-friendly machine learning tool for risk gene, cell type, and drug ranking for complex traits - GCDPipe. It uses gene-level GWAS-derived data and publicly available expression data to train a model for prediction of disease risk genes and relevant cell types. Gene-ranking information is then coupled with known drug targets data to prioritize drugs based on their estimated functional effects associated with identified risk genes. The pipeline was tested in two case studies: inflammatory bowel disease (IBD) and schizophrenia, then it was applied to Alzheimer's disease to investigate potential options for drug repurposing. The results show that GCDPipe is an effective tool to unify genetic risk factors with cellular context and known drug targets.