A deep learning method for predicting metabolite–disease associations via graph neural network

Abstract

Metabolism is the process by which an organism continuously replaces old substances with new substances. It plays an important role in maintaining human life, body growth and reproduction. More and more researchers have shown that the concentrations of some metabolites in patients are different from those in healthy people. Traditional biological experiments can test some hypotheses and verify their relationships but usually take a considerable amount of time and money. Therefore, it is urgent to develop a new computational method to identify the relationships between metabolites and diseases. In this work, we present a new deep learning algorithm named as graph convolutional network with graph attention network (GCNAT) to predict the potential associations of disease-related metabolites. First, we construct a heterogeneous network based on known metabolite–disease associations, metabolite–metabolite similarities and disease–disease similarities. Metabolite and disease features are encoded and learned through the graph convolutional neural network. Then, a graph attention layer is used to combine the embeddings of multiple convolutional layers, and the corresponding attention coefficients are calculated to assign different weights to the embeddings of each layer. Further, the prediction result is obtained by decoding and scoring the final synthetic embeddings. Finally, GCNAT achieves a reliable area under the receiver operating characteristic curve of 0.95 and the precision-recall curve of 0.405, which are better than the results of existing five state-of-the-art predictive methods in 5-fold cross-validation, and the case studies show that the metabolite–disease correlations predicted by our method can be successfully demonstrated by relevant experiments. We hope that GCNAT could be a useful biomedical research tool for predicting potential metabolite–disease associations in the future.

A deep ensemble learning-based automated detection of COVID-19 using lung CT images and Vision Transformer and ConvNeXt

“…However, during the initial stage of the epidemic outbreak, the massive influx of patients often means medical staff and healthcare professionals have to work 24 h a day, which has a bad effect on the physical and mental health of doctors and affects the accuracy and efficiency of the medical diagnosis (Zhan et al, 2021). Alternatively, artificial intelligence technology is a quite efficient strategy and obtains wide application in various fields (Chen et al, 2019;Liu et al, 2021aLiu et al, , 2022aTang et al, 2021;Wang et al, 2021;Zhang et al, 2021;Liang et al, 2022;Sun et al, 2022;Yang et al, 2022), and can be used to complement the work of radiologists. It can efficiently assist medical staff in judging symptoms, for example, pre-classifying pathological images or predicting sampling results, and thus can greatly reduce their working intensity.…”

Prioritizing potential circRNA biomarkers for bladder cancer and bladder urothelial cancer based on an ensemble model

“…Moreover, its prediction performance needs further improvement. In the future, we will consider biological features of circRNAs and develop more efficient machine learning, Frontiers in Genetics frontiersin.org especially ensemble learning models (Zhou et al, 2021a;Peng et al, 2022a) and deep learning models (Peng et al, 2021b;Zhou et al, 2021b;Sun et al, 2022;Yang et al, 2022) to discover potential biomarkers for bladder cancer and bladder urothelial cancer.…”

JSCSNCP-LMA: a method for predicting the association of lncRNA–miRNA

“…It not only has a good performance, but also has good expansibility. For example, metabolite-disease association prediction 91 , miRNA-disease association prediction 92 , lncRNA-disease association prediction 93 , and lncRNA-protein association prediction 94 .…”