Factored QTL analysis

We develop Factored QTL analysis (fqtl), a multivariate, sparse multiple regression model to jointly identify causal eQTLs and tissues of action. The key idea of our method is to assume that the matrix of effect sizes is low rank and sparse, and use a novel Bayesian prior to estimate the posterior distribution of that matrix.

• TWAS pipeline
• AD brain-specific TWAS analysis
• Tissue-specific TWAS analysis

Technical details:

• fQTL models
• Dense factors