Risk stratification and pathway analysis based on graph neural network and interpretable algorithm

Abstract

Background Pathway-based analysis of transcriptomic data has shown greater stability and better performance than traditional gene-based analysis. Until now, some pathway-based deep learning models have been developed for bioinformatic analysis, but these models have not fully considered the topological features of pathways, which limits the performance of the final prediction result. Results To address this issue, we propose a novel model, called PathGNN, which constructs a Graph Neural Networks (GNNs) model that can capture topological features of pathways. As a case, PathGNN was applied to predict long-term survival of four types of cancer and achieved promising predictive performance when compared to other common methods. Furthermore, the adoption of an interpretation algorithm enabled the identification of plausible pathways associated with survival. Conclusion PathGNN demonstrates that GNN can be effectively applied to build a pathway-based model, resulting in promising predictive power.

KNeXT: a NetworkX-based topologically relevant KEGG parser

“…We do not know of any automated parsers that modify or interpret equivalent identifiers to create isolated connected components. KNeXT is intended to bridge this gap and our results provide a comparative analysis with a modern widely-used parser called ( Sales et al, 2012 ), which has been cited in several recent studies ( Bianco et al, 2017 ; Gouy et al, 2017 ; Benedetti et al, 2020 ; Rahat et al, 2020 ; Hellstern et al, 2021 ; Liang et al, 2022 ).…”

“…We do not know of any automated parsers that modify or interpret equivalent identifiers to create isolated connected components. KNeXT is intended to bridge this gap and our results provide a comparative analysis with a modern widely-used parser called graphite (Sales et al, 2012), which has been cited in several recent studies (Bianco et al, 2017;Gouy et al, 2017;Benedetti et al, 2020;Rahat et al, 2020;Hellstern et al, 2021;Liang et al, 2022). To illustrate the advantages of KNeXT we compare our novel approach to graphite using the Homo sapiens Rat Sarcoma (RAS) signaling pathway, a highly complex pathway with multiple effectors and features (Vojtek and Der, 1998).…”

Graph Neural Networks in Cancer and Oncology Research: Emerging and Future Trends

“…Gao et al [30] and Kim [31] extended the basic framework to model inter-patient groupings, "patient similarity networks", likewise achieving performance improvements in survival prediction on different cancer datasets. Liang et al [32] incorporated topological features of pathway representation of the transcriptomic data into the cancer survival prediction models for four cancers, taking advantage of the natural pathway-graph structure mapping. Again, prediction performance was superior to that of conventional ML/DL, with an added value of most predictive pathways' delineation.…”

Unbiased curriculum learning enhanced global-local graph neural network for protein thermodynamic stability prediction

“…Numerous studies have applied graph neural networks (GNNs) to biological problems, which includes protein design ( Ingraham et al 2019 ), feature representation learning ( Jing et al 2021 ), expression referring ( Yang et al 2020a , b ), relationship prediction ( Satorras et al 2021 ), survival gene path analysis ( Liang et al 2022 ), disease diagnosis ( Xing et al 2022 ), medical image analysis ( Huang et al 2022 ), and human action analysis ( Yan et al 2023 ). However, none of these works focus on point mutations.…”