PT  - JOURNAL ARTICLE
AU  - Majdandzic, Antonio
AU  - Rajesh, Chandana
AU  - Koo, Peter K.
TI  - Correcting gradient-based interpretations of deep neural networks for genomics
AID  - 10.1101/2022.04.29.490102
DP  - 2022 Jan 01
TA  - bioRxiv
PG  - 2022.04.29.490102
4099  - http://biorxiv.org/content/early/2022/08/23/2022.04.29.490102.short
4100  - http://biorxiv.org/content/early/2022/08/23/2022.04.29.490102.full
AB  - Post-hoc attribution methods are widely applied to provide insights into patterns learned by deep neural networks (DNNs). Despite their success in regulatory genomics, DNNs can learn arbitrary functions outside the probabilistic simplex that defines one-hot encoded DNA. This introduces a random gradient component that manifests as noise in attribution scores. Here we demonstrate the pervasiveness of off-simplex gradient noise for genomic DNNs and introduce a statistical correction that is effective at improving the interpretability of attribution methods.Competing Interest StatementThe authors have declared no competing interest.