layerUMAP: A tool for visualizing and understanding deep learning models in biological sequence classification using UMAP

Runyu Jing; Li Xue; Menglong Li; Lezheng Yu; Jiesi Luo

doi:10.1016/j.isci.2022.105530

layerUMAP: A tool for visualizing and understanding deep learning models in biological sequence classification using UMAP

iScience. 2022 Nov 7;25(12):105530. doi: 10.1016/j.isci.2022.105530. eCollection 2022 Dec 22.

Authors

Runyu Jing¹, Li Xue², Menglong Li³, Lezheng Yu⁴, Jiesi Luo⁵

Affiliations

¹ School of Cyber Science and Engineering, Sichuan University, Chengdu 610065, Sichuan, China.
² School of Public Health, Southwest Medical University, Luzhou 646000, Sichuan, China.
³ College of Chemistry, Sichuan University, Chengdu 610065, Sichuan, China.
⁴ School of Chemistry and Materials Science, Guizhou Education University, Guiyang 550018, Guizhou, China.
⁵ Department of Pharmacology, School of Pharmacy, Southwest Medical University, Luzhou 646000, Sichuan, China.

Abstract

Despite the impressive success of deep learning techniques in various types of classification and prediction tasks, interpreting these models and explaining their predictions are still major challenges. In this article, we present an easy-to-use command line tool capable of visualizing and analyzing alternative representations of biological observations learned by deep learning models. This new tool, namely, layerUMAP, integrates autoBioSeqpy software and the UMAP library to address learned high-level representations. An important advantage of the tool is that it provides an interactive option that enables users to visualize the outputs of hidden layers along the depth of the model. We use two different classes of examples to illustrate the potential power of layerUMAP, and the results demonstrate that layerUMAP can provide insightful visual feedback about models and further guide us to develop better models.

Keywords: Artificial intelligence applications; Bioinformatics; Genomics; Systems biology; Transcriptomics.