Learning Distance-Dependent Motif Interactions: An Interpretable CNN Model of Genomic Events

Thomas Quinn⁰, Dang Nguyen⁰, Phuoc Nguyen⁰, Sunil Gupta⁰, Svetha Venkatesh⁰
(0) Applied Artificial Intelligence Institute (A2I2)

Find me on Tues Nov 24th, 1:40-3pm AEDT in Remo, table 5

Abstract
In most biological studies, prediction is used primarily to validate the model; the real quest is to understand the underlying phenomenon. Therefore, interpretable deep models for biological studies are required. Here, we propose HyperXPair (the Hyper-parameter eXplainable Motif Pair framework) to model biological motifs and their distance-dependent context through explicitly interpretable parameters. This makes HyperXPair more than a decision-support tool; it is also a hypothesis-generating tool designed to advance knowledge in the field. We demonstrate the utility of our model by learning distance-dependent motif interactions for two biological problems: transcription initiation and RNA splicing.