SIGANEO: Similarity network with GAN enhancement for immunogenic neoepitope prediction
Target selection of the personalized cancer neoantigen vaccine, which is highly dependent on computational prediction algorithms, is crucial for its clinical efficacy. Due to the limited number of experimentally validated immunogenic neoepitopes as well as the complexity of neoantigens in eliciting...
| Main Authors: | , , , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
Elsevier
2023-01-01
|
| Series: | Computational and Structural Biotechnology Journal |
| Subjects: | |
| Online Access: | http://www.sciencedirect.com/science/article/pii/S2001037023004051 |
| _version_ | 1797384075214848000 |
|---|---|
| author | Yilin Ye Yiming Shen Jian Wang Dong Li Yu Zhu Zhao Zhao Youdong Pan Yi Wang Xing Liu Ji Wan |
| author_facet | Yilin Ye Yiming Shen Jian Wang Dong Li Yu Zhu Zhao Zhao Youdong Pan Yi Wang Xing Liu Ji Wan |
| author_sort | Yilin Ye |
| collection | DOAJ |
| description | Target selection of the personalized cancer neoantigen vaccine, which is highly dependent on computational prediction algorithms, is crucial for its clinical efficacy. Due to the limited number of experimentally validated immunogenic neoepitopes as well as the complexity of neoantigens in eliciting T cell response, the accuracy of neoepitope immunogenicity prediction methods requires persistent efforts for improvement. We present a deep learning framework for neoepitope immunogenicity prediction – SIGANEO by integrating GAN-like network with similarity network to address issues of missing values and limited data concerning neoantigen prediction. This framework exhibits superior performance over competing machine-learning-based neoantigen prediction algorithms over an independent test dataset from TESLA consortium. Particularly for the clinical setting of neoantigen vaccine where only the top 10 and 20 predictions are selected for vaccine production, SIGANEO achieves significantly better accuracy for predicting experimentally validated neoepitopes. Our work demonstrates that deep learning techniques can greatly boost the accuracy of target identification for cancer neoantigen vaccine. |
| first_indexed | 2024-03-08T21:30:15Z |
| format | Article |
| id | doaj.art-907326f0d7d94a22a148a7aaa5877e16 |
| institution | Directory Open Access Journal |
| issn | 2001-0370 |
| language | English |
| last_indexed | 2024-03-08T21:30:15Z |
| publishDate | 2023-01-01 |
| publisher | Elsevier |
| record_format | Article |
| series | Computational and Structural Biotechnology Journal |
| spelling | doaj.art-907326f0d7d94a22a148a7aaa5877e162023-12-21T07:32:26ZengElsevierComputational and Structural Biotechnology Journal2001-03702023-01-012155385543SIGANEO: Similarity network with GAN enhancement for immunogenic neoepitope predictionYilin Ye0Yiming Shen1Jian Wang2Dong Li3Yu Zhu4Zhao Zhao5Youdong Pan6Yi Wang7Xing Liu8Ji Wan9Shenzhen Neocura Biotechnology Co. Ltd., Shenzhen 518055, ChinaShenzhen Neocura Biotechnology Co. Ltd., Shenzhen 518055, ChinaShenzhen Neocura Biotechnology Co. Ltd., Shenzhen 518055, ChinaShenzhen Neocura Biotechnology Co. Ltd., Shenzhen 518055, ChinaShenzhen Neocura Biotechnology Co. Ltd., Shenzhen 518055, ChinaShenzhen Neocura Biotechnology Co. Ltd., Shenzhen 518055, ChinaShenzhen Neocura Biotechnology Co. Ltd., Shenzhen 518055, ChinaShenzhen Neocura Biotechnology Co. Ltd., Shenzhen 518055, ChinaThe Center for Microbes, Development and Health, Key Laboratory of Molecular Virology and Immunology, Institut Pasteur of Shanghai, Chinese Academy of Sciences, Shanghai 200031, ChinaShenzhen Neocura Biotechnology Co. Ltd., Shenzhen 518055, China; Corresponding author.Target selection of the personalized cancer neoantigen vaccine, which is highly dependent on computational prediction algorithms, is crucial for its clinical efficacy. Due to the limited number of experimentally validated immunogenic neoepitopes as well as the complexity of neoantigens in eliciting T cell response, the accuracy of neoepitope immunogenicity prediction methods requires persistent efforts for improvement. We present a deep learning framework for neoepitope immunogenicity prediction – SIGANEO by integrating GAN-like network with similarity network to address issues of missing values and limited data concerning neoantigen prediction. This framework exhibits superior performance over competing machine-learning-based neoantigen prediction algorithms over an independent test dataset from TESLA consortium. Particularly for the clinical setting of neoantigen vaccine where only the top 10 and 20 predictions are selected for vaccine production, SIGANEO achieves significantly better accuracy for predicting experimentally validated neoepitopes. Our work demonstrates that deep learning techniques can greatly boost the accuracy of target identification for cancer neoantigen vaccine.http://www.sciencedirect.com/science/article/pii/S2001037023004051Deep learningNeoepitopeNeoantigenImmunogenicityCancer immunotherapy |
| spellingShingle | Yilin Ye Yiming Shen Jian Wang Dong Li Yu Zhu Zhao Zhao Youdong Pan Yi Wang Xing Liu Ji Wan SIGANEO: Similarity network with GAN enhancement for immunogenic neoepitope prediction Computational and Structural Biotechnology Journal Deep learning Neoepitope Neoantigen Immunogenicity Cancer immunotherapy |
| title | SIGANEO: Similarity network with GAN enhancement for immunogenic neoepitope prediction |
| title_full | SIGANEO: Similarity network with GAN enhancement for immunogenic neoepitope prediction |
| title_fullStr | SIGANEO: Similarity network with GAN enhancement for immunogenic neoepitope prediction |
| title_full_unstemmed | SIGANEO: Similarity network with GAN enhancement for immunogenic neoepitope prediction |
| title_short | SIGANEO: Similarity network with GAN enhancement for immunogenic neoepitope prediction |
| title_sort | siganeo similarity network with gan enhancement for immunogenic neoepitope prediction |
| topic | Deep learning Neoepitope Neoantigen Immunogenicity Cancer immunotherapy |
| url | http://www.sciencedirect.com/science/article/pii/S2001037023004051 |
| work_keys_str_mv | AT yilinye siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction AT yimingshen siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction AT jianwang siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction AT dongli siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction AT yuzhu siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction AT zhaozhao siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction AT youdongpan siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction AT yiwang siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction AT xingliu siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction AT jiwan siganeosimilaritynetworkwithganenhancementforimmunogenicneoepitopeprediction |