Deep interpretability for GWAS

These effects Genome-Wide Association Studies are typically may play an important role in complex diseases (Bell conducted using linear models to find genetic variants et al., 2011). Since deep networks are known to be able to associated with common diseases. In these model arbitrarily complicated nonlinear functions of their studies, association testing is done on a variant-byvariant inputs (Goodfellow et al., 2016), we aim to use them to basis, possibly missing out on nonlinear predict complex disease outcomes from genetic data, and interaction effects between variants. Deep networks then interpret them (sample by sample) to discover novel can be used to model these interactions, but interactions between SNPs that are significant predictors of they are difficult to train and interpret on large the disease outcome.