Scholarlink

<p>With the continuous advancement of biomedical research, multifunctional peptides are playing an increasingly crucial role in the biomedical field and demonstrating tremendous potential in areas such as disease treatment and drug development. Nowadays, the efficient prediction of multifunctional therapeutic peptides is not only of vital importance but also highly challenging, as a vast amount of peptide sequence data urgently awaits in-depth analysis. Against this background, a novel computational model named PLMCCL-TP has emerged, aiming to predict multifunctional therapeutic peptides. PLMCCL-TP innovatively combines the deep feature learning of protein language models (PLM) with clustering spatial representation, contrastive learning, and multi-scale convolutional feature extraction. It makes a breakthrough in capturing the characteristics of multifunctional peptides and solving the multi-label classification problem, successfully applying PLM and contrastive learning to the field of multi-label recognition. By constructing the clustering space, we can make better use of the potential relationships among functional labels, thus enhancing the prediction accuracy. Meanwhile, due to the advantages of the multi-scale convolutional network in extracting features at different levels, the model can excellently extract the local and global feature information of sequences and exhibits remarkable robustness when dealing with the complexity of peptide sequences. Compared with the currently advanced methods, the experimental results strongly confirm that the method proposed in this paper outperforms the existing methods on the MFTP benchmark dataset. PLMCCL-TP is available at <a href="https://github.com/jiangkun-lin/PLMCCL-TP">https://github.com/jiangkun-lin/PLMCCL-TP</a>.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

PLMCCL-TP: The protein language model and clustering method based on contrastive learning applied to the multifunctional therapeutic peptide identification model

  • 期刊论文

  • 发表日期：

    2025年9月19日

查看全文
  • Jiangkun Lin,
  • Shouheng Tuo

摘要

With the continuous advancement of biomedical research, multifunctional peptides are playing an increasingly crucial role in the biomedical field and demonstrating tremendous potential in areas such as disease treatment and drug development. Nowadays, the efficient prediction of multifunctional therapeutic peptides is not only of vital importance but also highly challenging, as a vast amount of peptide sequence data urgently awaits in-depth analysis. Against this background, a novel computational model named PLMCCL-TP has emerged, aiming to predict multifunctional therapeutic peptides. PLMCCL-TP innovatively combines the deep feature learning of protein language models (PLM) with clustering spatial representation, contrastive learning, and multi-scale convolutional feature extraction. It makes a breakthrough in capturing the characteristics of multifunctional peptides and solving the multi-label classification problem, successfully applying PLM and contrastive learning to the field of multi-label recognition. By constructing the clustering space, we can make better use of the potential relationships among functional labels, thus enhancing the prediction accuracy. Meanwhile, due to the advantages of the multi-scale convolutional network in extracting features at different levels, the model can excellently extract the local and global feature information of sequences and exhibits remarkable robustness when dealing with the complexity of peptide sequences. Compared with the currently advanced methods, the experimental results strongly confirm that the method proposed in this paper outperforms the existing methods on the MFTP benchmark dataset. PLMCCL-TP is available at https://github.com/jiangkun-lin/PLMCCL-TP.