Introduction
Peptides are essential to life, and understanding their bioactive function can facilitate the application and mechanistic understanding of peptides. Through an enormous experimental effort, the function of portion peptides have been determined, but this represents a small fraction of the bioactive peptides. Accurate computational approaches are needed to address this gap. Despite recent progress, the majority of past research has focused on the mono-functional bioactive peptides prediction and existing methods fall short of multi-functional peptide prediction. In this study, we provide the multi-label framework using protein language model and graph attention network for identifying multi-functional peptides. Pretrained protein language model is used to extract fine-grained features of different functional peptides from peptide sequences. We transformed the multi-label problem to graph node classification and the graph attention network is used to learn the relationship between different labels. We conducted a series of experiments on two datasets of multi-functional bioactive peptides(MFBP) and multi-functional therapeutics peptides(MFTP) to validate our framework. On the MFBP test set, Our approach's accuracy and absolute truth are 0.796 and 0.788, respectively, with 6.9% and 8.4% greater than the suboptimal method. On the MFTP test set, Accuracy and Absolute true of our method are 0.689 and 0.616, with 3.8% and 2.3% higher than those of the best historical model, respectively. The result demonstrates our model greatly outperforms other methods and we believe it can be effectively expanded to other multi-label studies in bioinformatics.
Citation
Upon the usage of this server the users are requested to use the following citation:
Luo, J., Zhao, K., Chen, J., Yang, C., Qu, F., Liu, Y., Jin, X., Yan, K., Zhang, Y. and Liu, B.iMFP-LG: Identification of Novel Multi-Functional Peptides by Using Protein Language Models and Graph-Based Deep Learning. Genomics, Proteomics & Bioinformatics, 2024: qzae084.
References
1. Tang, W., Dai, R., Yan, W., Zhang, W., Bin, Y., Xia, E., & Xia, J. (2022). Identifying multi-functional bioactive peptide functions using multi-label deep learning. Briefings in Bioinformatics, 23(1), bbab414.
2. Yan, W., Tang, W., Wang, L., Bin, Y., & Xia, J. (2022). PrMFTP: Multi-functional therapeutic peptides prediction based on multi-head self-attention mechanism and class weight optimization. PLoS Computational Biology, 18(9), e1010511.
3. Li, Y., Li, X., Liu, Y., Yao, Y., & Huang, G. (2022). MPMABP: a CNN and Bi-LSTM-Based method for predicting multi-activities of bioactive peptides. Pharmaceuticals, 15(6), 707.