RiVIERA-beta: Joint Bayesian inference of risk variants and tissue-specific epigenomic enrichments across multiple complex human diseases

Genome wide association studies (GWAS) provide a powerful approach for un11 covering disease-associated variants in human, but fine-mapping the causal variants 12 remains a challenge. This is partly remedied by prioritization of disease-associated vari13 ants that overlap GWAS-enriched epigenomic annotations. Here, we introduce a new 14 Bayesian model RiVIERA-beta (Risk Variant Inference using Epigenomic Reference 15 Annotations) for inference of driver variants by modelling summary statistics p-values 16 in Beta density function across multiple traits using hundreds of epigenomic annota17 tions. In simulation, RiVIERA-beta promising power in detecting causal variants and 18 causal annotations, the multi-trait joint inference further improved the detection power. 19 We applied RiVIERA-beta to model the existing GWAS summary statistics of 9 au20 toimmune diseases and Schizophrenia by jointly harnessing the potential causal enrich21 ments among 848 tissue-specific epigenomics annotations from ENCODE/Roadmap 22 consortium covering 127 cell/tissue types and 8 major epigenomic marks. RiVIERA23 beta identified meaningful tissue-specific enrichments for enhancer regions defined by 24 H3K4me1 and H3K27ac for Blood T-Cell specifically in the 9 autoimmune diseases 25 and Brain-specific enhancer activities exclusively in Schizophrenia. Moreover, the 26 variants from the 95% credible sets exhibited high conservation and enrichments for 27 GTEx whole-blood eQTLs located within transcription-factor-binding-sites and DNA28 hypersensitive-sites. Furthermore, joint modeling the nine immune traits by simulta29 neously inferring and exploiting the underlying epigenomic correlation between traits 30 further improved the functional enrichments compared to single-trait models. 31