Shaylyn Govender, Emily Morgan, Rabelani Ramahala, Kevin Lobb, Nigel T Bishop, Özlem Tastan Bishop
{"title":"Transfer learning towards predicting viral missense mutations: A case study on SARS-CoV-2.","authors":"Shaylyn Govender, Emily Morgan, Rabelani Ramahala, Kevin Lobb, Nigel T Bishop, Özlem Tastan Bishop","doi":"10.1016/j.csbj.2025.04.029","DOIUrl":null,"url":null,"abstract":"<p><p>Understanding viral evolution and predicting future mutations are crucial for overcoming drug resistance and developing long-lasting treatments. Previously, we established machine learning (ML) models using dynamic residue network (DRN) metric data and leveraging a vast amount of existing mutation data from the SARS-CoV-2 main protease (M<sup>pro</sup>). Here, we sought to assess the generalizability and robustness of the current models across other SARS-CoV-2 proteins. To achieve this, for the first time, we employed a transfer learning (TL) approach, allowing us to determine the extent to which M<sup>pro</sup> trained models could be applied to other SARS-CoV-2 proteins. The TL results were highly promising, with artificial neural network (ANN) and random forest (RF) correlation coefficients for M<sup>pro</sup> closely matching those of NSP10, NSP16, and PL<sup>pro</sup>. The ANN |R| value for M<sup>pro</sup> was 0.564, while NSP10, NSP16, and PL<sup>pro</sup> had values of 0.533, 0.527, and 0.464, respectively. Similarly, the RF |R| value for M<sup>pro</sup> was 0.673, compared to 0.457, 0.460, and 0.437 for NSP10, NSP16, and PL<sup>pro</sup>, respectively. Interestingly, we did not observe a strong correlation for the spike (S) protein monomer and its domains. The low p-values that are associated with the correlation |R| values show that the linear correlations between predicted and actual mutation frequencies are statistically significant. This indicates that TL may generalize well across structurally related viral proteins using DRN-derived ML model from M<sup>pro</sup>. Overall, we aim to develop a universal ML model for predicting missense mutation frequencies in viral proteins, and this study lays the foundation for that goal.</p>","PeriodicalId":10715,"journal":{"name":"Computational and structural biotechnology journal","volume":"27 ","pages":"1686-1692"},"PeriodicalIF":4.4000,"publicationDate":"2025-04-22","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12063013/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Computational and structural biotechnology journal","FirstCategoryId":"99","ListUrlMain":"https://doi.org/10.1016/j.csbj.2025.04.029","RegionNum":2,"RegionCategory":"生物学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"2025/1/1 0:00:00","PubModel":"eCollection","JCR":"Q2","JCRName":"BIOCHEMISTRY & MOLECULAR BIOLOGY","Score":null,"Total":0}
引用次数: 0
Abstract
Understanding viral evolution and predicting future mutations are crucial for overcoming drug resistance and developing long-lasting treatments. Previously, we established machine learning (ML) models using dynamic residue network (DRN) metric data and leveraging a vast amount of existing mutation data from the SARS-CoV-2 main protease (Mpro). Here, we sought to assess the generalizability and robustness of the current models across other SARS-CoV-2 proteins. To achieve this, for the first time, we employed a transfer learning (TL) approach, allowing us to determine the extent to which Mpro trained models could be applied to other SARS-CoV-2 proteins. The TL results were highly promising, with artificial neural network (ANN) and random forest (RF) correlation coefficients for Mpro closely matching those of NSP10, NSP16, and PLpro. The ANN |R| value for Mpro was 0.564, while NSP10, NSP16, and PLpro had values of 0.533, 0.527, and 0.464, respectively. Similarly, the RF |R| value for Mpro was 0.673, compared to 0.457, 0.460, and 0.437 for NSP10, NSP16, and PLpro, respectively. Interestingly, we did not observe a strong correlation for the spike (S) protein monomer and its domains. The low p-values that are associated with the correlation |R| values show that the linear correlations between predicted and actual mutation frequencies are statistically significant. This indicates that TL may generalize well across structurally related viral proteins using DRN-derived ML model from Mpro. Overall, we aim to develop a universal ML model for predicting missense mutation frequencies in viral proteins, and this study lays the foundation for that goal.
期刊介绍:
Computational and Structural Biotechnology Journal (CSBJ) is an online gold open access journal publishing research articles and reviews after full peer review. All articles are published, without barriers to access, immediately upon acceptance. The journal places a strong emphasis on functional and mechanistic understanding of how molecular components in a biological process work together through the application of computational methods. Structural data may provide such insights, but they are not a pre-requisite for publication in the journal. Specific areas of interest include, but are not limited to:
Structure and function of proteins, nucleic acids and other macromolecules
Structure and function of multi-component complexes
Protein folding, processing and degradation
Enzymology
Computational and structural studies of plant systems
Microbial Informatics
Genomics
Proteomics
Metabolomics
Algorithms and Hypothesis in Bioinformatics
Mathematical and Theoretical Biology
Computational Chemistry and Drug Discovery
Microscopy and Molecular Imaging
Nanotechnology
Systems and Synthetic Biology