M Torabi, I Haririan, A Foroumadi, H Ghanbari, F Ghasemi
{"title":"基于 BERT 预训练模型的深度学习模型,用于预测抗癌化学物质的抗增殖活性。","authors":"M Torabi, I Haririan, A Foroumadi, H Ghanbari, F Ghasemi","doi":"10.1080/1062936X.2024.2431486","DOIUrl":null,"url":null,"abstract":"<p><p>Identifying new compounds with minimal side effects to enhance patients' quality of life is the ultimate goal of drug discovery. Due to the expensive and time-consuming nature of experimental investigations and the scarcity of data in traditional QSAR studies, deep transfer learning models, such as the BERT model, have recently been suggested. This study evaluated the model's performance in predicting the anti-proliferative activity of five cancer cell lines (HeLa, MCF7, MDA-MB231, PC3, and MDA-MB) using over 3,000 synthesized molecules from PubChem. The results indicated that the model could predict the class of designed small molecules with acceptable accuracy for most cell lines, except for PC3 and MDA-MB. The model's performance was further tested on an in-house dataset of approximately 25 small molecules per cell line, based on IC50 values. The model accurately predicted the biological activity class for HeLa with an accuracy of <math><mn>0.77</mn><mo>±</mo><mn>0.4</mn></math> and demonstrated acceptable performance for MCF7 and MDA-MB231, with accuracy between 0.56 and 0.66. However, the results were less reliable for PC3 and HepG2. In conclusion, the ChemBERTa fine-tuned model shows potential for predicting outcomes on in-house datasets.</p>","PeriodicalId":21446,"journal":{"name":"SAR and QSAR in Environmental Research","volume":" ","pages":"1-22"},"PeriodicalIF":2.3000,"publicationDate":"2024-11-28","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":"{\"title\":\"A deep learning model based on the BERT pre-trained model to predict the antiproliferative activity of anti-cancer chemical compounds.\",\"authors\":\"M Torabi, I Haririan, A Foroumadi, H Ghanbari, F Ghasemi\",\"doi\":\"10.1080/1062936X.2024.2431486\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<p><p>Identifying new compounds with minimal side effects to enhance patients' quality of life is the ultimate goal of drug discovery. Due to the expensive and time-consuming nature of experimental investigations and the scarcity of data in traditional QSAR studies, deep transfer learning models, such as the BERT model, have recently been suggested. This study evaluated the model's performance in predicting the anti-proliferative activity of five cancer cell lines (HeLa, MCF7, MDA-MB231, PC3, and MDA-MB) using over 3,000 synthesized molecules from PubChem. The results indicated that the model could predict the class of designed small molecules with acceptable accuracy for most cell lines, except for PC3 and MDA-MB. The model's performance was further tested on an in-house dataset of approximately 25 small molecules per cell line, based on IC50 values. The model accurately predicted the biological activity class for HeLa with an accuracy of <math><mn>0.77</mn><mo>±</mo><mn>0.4</mn></math> and demonstrated acceptable performance for MCF7 and MDA-MB231, with accuracy between 0.56 and 0.66. However, the results were less reliable for PC3 and HepG2. In conclusion, the ChemBERTa fine-tuned model shows potential for predicting outcomes on in-house datasets.</p>\",\"PeriodicalId\":21446,\"journal\":{\"name\":\"SAR and QSAR in Environmental Research\",\"volume\":\" \",\"pages\":\"1-22\"},\"PeriodicalIF\":2.3000,\"publicationDate\":\"2024-11-28\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"SAR and QSAR in Environmental Research\",\"FirstCategoryId\":\"93\",\"ListUrlMain\":\"https://doi.org/10.1080/1062936X.2024.2431486\",\"RegionNum\":3,\"RegionCategory\":\"环境科学与生态学\",\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q3\",\"JCRName\":\"CHEMISTRY, MULTIDISCIPLINARY\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"SAR and QSAR in Environmental Research","FirstCategoryId":"93","ListUrlMain":"https://doi.org/10.1080/1062936X.2024.2431486","RegionNum":3,"RegionCategory":"环境科学与生态学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q3","JCRName":"CHEMISTRY, MULTIDISCIPLINARY","Score":null,"Total":0}
A deep learning model based on the BERT pre-trained model to predict the antiproliferative activity of anti-cancer chemical compounds.
Identifying new compounds with minimal side effects to enhance patients' quality of life is the ultimate goal of drug discovery. Due to the expensive and time-consuming nature of experimental investigations and the scarcity of data in traditional QSAR studies, deep transfer learning models, such as the BERT model, have recently been suggested. This study evaluated the model's performance in predicting the anti-proliferative activity of five cancer cell lines (HeLa, MCF7, MDA-MB231, PC3, and MDA-MB) using over 3,000 synthesized molecules from PubChem. The results indicated that the model could predict the class of designed small molecules with acceptable accuracy for most cell lines, except for PC3 and MDA-MB. The model's performance was further tested on an in-house dataset of approximately 25 small molecules per cell line, based on IC50 values. The model accurately predicted the biological activity class for HeLa with an accuracy of and demonstrated acceptable performance for MCF7 and MDA-MB231, with accuracy between 0.56 and 0.66. However, the results were less reliable for PC3 and HepG2. In conclusion, the ChemBERTa fine-tuned model shows potential for predicting outcomes on in-house datasets.
期刊介绍:
SAR and QSAR in Environmental Research is an international journal welcoming papers on the fundamental and practical aspects of the structure-activity and structure-property relationships in the fields of environmental science, agrochemistry, toxicology, pharmacology and applied chemistry. A unique aspect of the journal is the focus on emerging techniques for the building of SAR and QSAR models in these widely varying fields. The scope of the journal includes, but is not limited to, the topics of topological and physicochemical descriptors, mathematical, statistical and graphical methods for data analysis, computer methods and programs, original applications and comparative studies. In addition to primary scientific papers, the journal contains reviews of books and software and news of conferences. Special issues on topics of current and widespread interest to the SAR and QSAR community will be published from time to time.