Software Open Access

Biociphers/eig: Accepted Manuscript - Genome Biology

Jha, Anupama; Aicher, Joseph K.; Gazzara, Matthew R.; Singh, Deependra; Barash, Yoseph

Publication: Enhanced Integrated Gradients: improving interpretability of deep learning models using splicing codes as a case study, accepted to Genome Biology 

This repository reproduces results from the EIG publication. To use EIG with your own deep learning model, please refer to the repository (https://bitbucket.org/biociphers/eig) which contains instructions to install the EIG package and example usage.

Abstract

Despite the success and fast adaptation of deep learning models in biomedical domains, their lack of interpretability remains an issue. Here we introduce Enhanced Integrated Gradients (EIG), a method to identify significant features associated with a specific prediction task. Using RNA splicing prediction as well as digit classification as case studies, we demonstrate that EIG improves upon the original Integrated Gradients method and produces sets of informative features. We then apply EIG to identify A1CF as a key regulator of liver specific alternative splicing, supporting this finding with subsequent analysis of relevant A1CF functional (RNA-seq) and binding data (PAR-CLIP).

Files (1.5 MB)
Name Size
jha_eig_2020_reproducibility.zip
md5:8994d82ca6cc0d1a804300ab82bb1d8b
1.5 MB Download
96
7
views
downloads
All versions This version
Views 9675
Downloads 75
Data volume 17.4 MB7.4 MB
Unique views 7567
Unique downloads 65

Share

Cite as