Understanding the genetic basis of phenotypic variation is fundamental to biology. Here we introduce GAP, a novel machine learning framework for predicting binary phenotypes from gaps in multi-species sequence alignments. GAP employs a neural network...