Heewon Park ,Koji Maruhashi,Rui Yamaguchi,Seiya Imoto,Satoru Miyano
Abstract
In recent years, personalized gene regulatory networks have received significant attention, and interpretation of the multilayer networks has been a critical issue for a comprehensive understanding of gene regulatory systems. Although several statistical and machine learning approaches have been developed and applied to reveal sample-specific regulatory pathways, integrative understanding of the massive multilayer networks remains a challenge. To resolve this problem, we propose a novel artificial intelligence (AI) strategy for comprehensive gene regulatory network analysis. In our strategy, personalized gene networks corresponding specific clinical characteristic are constructed and the constructed network is considered as a second-order tensor. Then, an explainable AI method based on deep learning is applied to decompose the multilayer networks, thus we can reveal all-encompassing gene regulatory systems characterized by clinical features of patients. To evaluate the proposed methodology, we apply our method to the multilayer gene networks under varying conditions of an epithelial–mesenchymal transition (EMT) process. From the comprehensive analysis of multilayer networks, we identified novel markers, and the biological mechanisms of the identified genes and their reciprocal mechanisms are verified through the literature. Although any biological knowledge about the identified genes was not incorporated in our analysis, our data-driven approach based on AI approach provides biologically reliable results. Furthermore, the results provide crucial evidences to reveal biological mechanism related to various diseases, e.g., keratinocyte proliferation. The use of explainable AI method based on the tensor decomposition enables us to reveal global and novel mechanisms of gene regulatory system from the massive multiple networks, which cannot be demonstrated by existing methods. We expect that the proposed method provides a new insight into network biology and it will be a useful tool to integrative gene network analysis related complex architectures of diseases.
Introduction
Gene regulatory networks are crucial for understanding complex mechanisms of diseases. To reveal heterogeneous genetic networks that underlie complex human diseases, various large-scale projects (e.g., The Cancer Genome Atlas and Cancer Genome Project) have been conducted and provided considerable amounts of omics data. The scale of gene networks is increasing, and strategies to comprehensively analyze large-scale gene networks have been claimed. In particular, there is currently substantial discussion regarding integrative analysis of sample-specific gene networks for personalized cancer diagnostics and therapeutics.
Methods
Suppose X1, …, Xq is q possible regulators that may control transcription of the lth target gene Yl. Consider the linear regression model for the target gene Yl,
Discussion
We introduced a novel methodology for a comprehensive analysis of large-scale personalized network tensors. In this study, we considered a gene regulatory network as a tensor for a data point, and decompose the multilayer networks represented as tensors by using an AI approach, TRIP. Unlike existing studies for sample-specific gene network construction, our strategy analyzes whole multilayer networks based on tensor decomposition, thus we can perform wide exploration of the large-scale gene regulatory networks for all patients. To illustrate our method, we applied it to personalized networks constructed for 762 cell lines having varying conditions of the EMT process. We identified novel candidate markers and verified biological mechanisms of a majority of the identified markers based on the literature. Although most of the identified markers were found in previous studies, some of the revealed genes could not be verified. Further work is required towards experimental validation of the newly revealed markers. In this study, our strategy was illustrated based only on EMT-related gene regulatory networks. As one of our future works, we consider the comprehensive analysis of dynamic systems of personalized gene networks in accordance with various clinical characteristics (e.g., drug sensitivity of cell lines).
Acknowledgments
The authors thank the associate editor and anonymous reviewers for the constructive and valuable comments that improved the quality of the paper considerably. This research used computational resources of the Super Computer System, Human Genome Center, Institute of Medical Science, University of Tokyo.
Citation: Park H, Maruhashi K, Yamaguchi R, Imoto S, Miyano S (2020) Global gene network exploration based on explainable artificial intelligence approach. PLoS ONE 15(11): e0241508. https://doi.org/10.1371/journal.pone.0241508
Editor: Gary Stein, University of Vermont, UNITED STATES
Received: July 18, 2020; Accepted: September 3, 2020; Published: November 6, 2020.
Copyright: © 2020 Park et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability:
Data are available from: https://www.kaggle.com/heewonn/poned2022295.
Funding:
The authors received no specific funding for this work.
Competing interests:
The authors have declared that no competing interests exist.
