Drug target prediction through deep learning functional representation of gene signatures
Abstract Many machine learning applications in bioinformatics currently rely on matching gene identities when analyzing input gene signatures and fail to take advantage of preexisting knowledge about gene functions. To further enable comparative analysis of OMICS datasets, including target deconvolu...
| Main Authors: | , , , , , , , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Nature Portfolio
2024-02-01
|
| Series: | Nature Communications |
| Online Access: | https://doi.org/10.1038/s41467-024-46089-y |
| _version_ | 1797274017559740416 |
|---|---|
| author | Hao Chen Frederick J. King Bin Zhou Yu Wang Carter J. Canedy Joel Hayashi Yang Zhong Max W. Chang Lars Pache Julian L. Wong Yong Jia John Joslin Tao Jiang Christopher Benner Sumit K. Chanda Yingyao Zhou |
| author_facet | Hao Chen Frederick J. King Bin Zhou Yu Wang Carter J. Canedy Joel Hayashi Yang Zhong Max W. Chang Lars Pache Julian L. Wong Yong Jia John Joslin Tao Jiang Christopher Benner Sumit K. Chanda Yingyao Zhou |
| author_sort | Hao Chen |
| collection | DOAJ |
| description | Abstract Many machine learning applications in bioinformatics currently rely on matching gene identities when analyzing input gene signatures and fail to take advantage of preexisting knowledge about gene functions. To further enable comparative analysis of OMICS datasets, including target deconvolution and mechanism of action studies, we develop an approach that represents gene signatures projected onto their biological functions, instead of their identities, similar to how the word2vec technique works in natural language processing. We develop the Functional Representation of Gene Signatures (FRoGS) approach by training a deep learning model and demonstrate that its application to the Broad Institute’s L1000 datasets results in more effective compound-target predictions than models based on gene identities alone. By integrating additional pharmacological activity data sources, FRoGS significantly increases the number of high-quality compound-target predictions relative to existing approaches, many of which are supported by in silico and/or experimental evidence. These results underscore the general utility of FRoGS in machine learning-based bioinformatics applications. Prediction networks pre-equipped with the knowledge of gene functions may help uncover new relationships among gene signatures acquired by large-scale OMICs studies on compounds, cell types, disease models, and patient cohorts. |
| first_indexed | 2024-03-07T14:52:23Z |
| format | Article |
| id | doaj.art-3b4f238e44764f15bb18667e531420c4 |
| institution | Directory Open Access Journal |
| issn | 2041-1723 |
| language | English |
| last_indexed | 2024-03-07T14:52:23Z |
| publishDate | 2024-02-01 |
| publisher | Nature Portfolio |
| record_format | Article |
| series | Nature Communications |
| spelling | doaj.art-3b4f238e44764f15bb18667e531420c42024-03-05T19:37:51ZengNature PortfolioNature Communications2041-17232024-02-0115111510.1038/s41467-024-46089-yDrug target prediction through deep learning functional representation of gene signaturesHao Chen0Frederick J. King1Bin Zhou2Yu Wang3Carter J. Canedy4Joel Hayashi5Yang Zhong6Max W. Chang7Lars Pache8Julian L. Wong9Yong Jia10John Joslin11Tao Jiang12Christopher Benner13Sumit K. Chanda14Yingyao Zhou15Novartis Biomedical ResearchNovartis Biomedical ResearchNovartis Biomedical ResearchNovartis Biomedical ResearchNovartis Biomedical ResearchNovartis Biomedical ResearchNovartis Biomedical ResearchDepartment of Medicine, University of California, San DiegoNCI Designated Cancer Center, Sanford Burnham Prebys Medical Discovery InstituteNovartis Biomedical ResearchNovartis Biomedical ResearchNovartis Biomedical ResearchDepartment of Computer Science and Engineering, University of California, RiversideDepartment of Medicine, University of California, San DiegoDepartment of Immunology and Microbiology, Scripps ResearchNovartis Biomedical ResearchAbstract Many machine learning applications in bioinformatics currently rely on matching gene identities when analyzing input gene signatures and fail to take advantage of preexisting knowledge about gene functions. To further enable comparative analysis of OMICS datasets, including target deconvolution and mechanism of action studies, we develop an approach that represents gene signatures projected onto their biological functions, instead of their identities, similar to how the word2vec technique works in natural language processing. We develop the Functional Representation of Gene Signatures (FRoGS) approach by training a deep learning model and demonstrate that its application to the Broad Institute’s L1000 datasets results in more effective compound-target predictions than models based on gene identities alone. By integrating additional pharmacological activity data sources, FRoGS significantly increases the number of high-quality compound-target predictions relative to existing approaches, many of which are supported by in silico and/or experimental evidence. These results underscore the general utility of FRoGS in machine learning-based bioinformatics applications. Prediction networks pre-equipped with the knowledge of gene functions may help uncover new relationships among gene signatures acquired by large-scale OMICs studies on compounds, cell types, disease models, and patient cohorts.https://doi.org/10.1038/s41467-024-46089-y |
| spellingShingle | Hao Chen Frederick J. King Bin Zhou Yu Wang Carter J. Canedy Joel Hayashi Yang Zhong Max W. Chang Lars Pache Julian L. Wong Yong Jia John Joslin Tao Jiang Christopher Benner Sumit K. Chanda Yingyao Zhou Drug target prediction through deep learning functional representation of gene signatures Nature Communications |
| title | Drug target prediction through deep learning functional representation of gene signatures |
| title_full | Drug target prediction through deep learning functional representation of gene signatures |
| title_fullStr | Drug target prediction through deep learning functional representation of gene signatures |
| title_full_unstemmed | Drug target prediction through deep learning functional representation of gene signatures |
| title_short | Drug target prediction through deep learning functional representation of gene signatures |
| title_sort | drug target prediction through deep learning functional representation of gene signatures |
| url | https://doi.org/10.1038/s41467-024-46089-y |
| work_keys_str_mv | AT haochen drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT frederickjking drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT binzhou drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT yuwang drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT carterjcanedy drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT joelhayashi drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT yangzhong drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT maxwchang drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT larspache drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT julianlwong drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT yongjia drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT johnjoslin drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT taojiang drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT christopherbenner drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT sumitkchanda drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures AT yingyaozhou drugtargetpredictionthroughdeeplearningfunctionalrepresentationofgenesignatures |