In summary, we have presented a statistical approach, named GPA, that can integrate information from multiple GWAS data sets and functional annotation data. Not only does GPA have better statistical power than related methods, it also provides interpretable model parameters offering insights to our understanding of the genetic architecture of complex traits. We have successfully applied GPA to analyze GWAS data of five psychiatric disorders from PGC, and showed that GPA is able to identify pleiotropic effects among psychiatric disorders and detect enrichment of the CNS gene set. We have also applied GPA to analyze a bladder cancer GWAS dataset with ENCODE data as annotation, where significant enrichments of immune system and carcinoma pathways were observed. Compared to LMM that requires individual genotype and phenotype data as input, GPA has similar results of enrichment analysis without requirements of the genotype data. This makes GPA an attractive and effective tool for the integrative analysis of multiple GWAS data with functional annotation data, when genotype data are not available.
