Deep learning models capture nonlinear genotype-phenotype mappings

Multilayer perceptrons and related architectures are able to approximate nonlinear functions, enabling them to model epistatic interactions in genotype-phenotype datasets where linear regression fails.

Part of Chain