iKcr-RG:A Two-branching Strategy Based on ResNet and BiGRU to Predict Lysine Acylation Sites of Non-histone Proteins
10.13865/j.cnki.cjbmb.2025.01.1354
- VernacularTitle:iKcr-RG:基于ResNet和BiGRU的双分支策略预测非组蛋白质赖氨酸酰化位点
- Author:
Hang CHENG
1
;
Yan-Bei DUAN
;
Xin WEI
Author Information
1. 南昌工学院教务处,南昌 330108
- Publication Type:Journal Article
- Keywords:
protein post-translational modification(PTM);
lysine acylation site;
deep learning;
feature extraction;
feature fusion
- From:
Chinese Journal of Biochemistry and Molecular Biology
2025;41(2):305-314
- CountryChina
- Language:Chinese
-
Abstract:
Lyase acylation is a post-translational modification of proteins that plays a key role in cellular function,gene transcription and cellular metabolism.Meanwhile,lysine acylation is also involved in mul-tiple biological processes in life forms,and its abnormality may be associated with the occurrence and de-velopment of many diseases.Therefore,prediction of lysine acylation sites is important for the diagnosis and treatment of diseases.Although biomedical experiments can detect lysine acylation sites with high precision,they are costly and time-consuming.To address this problem,researchers have developed more convenient and efficient computational methods as an alternative to traditional biomedical experi-mental techniques.In this study,we developed a prediction model iKcr-RG based on a deep learning ap-proach,which employs a two-branching strategy using both ResNet and BiGRU to extract both local and global feature information from the original sequence encoding.To further improve the performance of the model,we innovatively designed a feature fusion method.After these optimizations,this study demon-strates stronger robustness and stability in unbalanced data.In the independent dataset results,specificity(Sp),sensitivity(Sn),accuracy(Acc)and Matthews correlation coefficient(MCC)were 0.8109,0.7902,0.7940,and 0.4978 respectively.The iKcr-RG model was better than existing prediction mod-els in predicting lysine acylation sites,and this study provides new ideas and methods for the application of deep learning in bioinformatics.
