PSSP-MVIRT: peptide secondary structure prediction based on a multi-view deep learning architecture

Abstract

The prediction of peptide secondary structures is fundamentally important to reveal the functional mechanisms of peptides with potential applications as therapeutic molecules. In this study, we propose a multi-view deep learning method named Peptide Secondary Structure Prediction based on Multi-View Information, Restriction and Transfer learning (PSSP-MVIRT) for peptide secondary structure prediction. To sufficiently exploit discriminative information, we introduce a multi-view fusion strategy to integrate different information from multiple perspectives, including sequential information, evolutionary information and hidden state information, respectively, and generate a unified feature space. Moreover, we construct a hybrid network architecture of Convolutional Neural Network and Bi-directional Gated Recurrent Unit to extract global and local features of peptides. Furthermore, we utilize transfer learning to effectively alleviate the lack of training samples (peptides with experimentally validated structures). Comparative results on independent tests demonstrate that our proposed method significantly outperforms state-of-the-art methods. In particular, our method exhibits better performance at the segment level, suggesting the strong ability of our model in capturing local discriminative information. The case study also shows that our PSSP-MVIRT achieves promising and robust performance in the prediction of new peptide secondary structures. Importantly, we establish a webserver to implement the proposed method, which is currently accessible via http://server.malab.cn/PSSP-MVIRT. We expect it can be a useful tool for the researchers of interest, facilitating the wide use of our method.

Explainable deep graph learning accurately modeling the peptide secondary structure prediction

“…Previous studies have demonstrated that the functionality of peptides (e.g., affinity) is easily affected by the length of sequences, with most bioactive peptides being normally less than 40 residues long[19, 38, 39]. To investigate if our model had length biases for peptide secondary structure prediction, we further explored whether peptide length affected the performance of our model.…”

“…To evaluate the performance of the proposed PHAT model, we compared it with four state-of-the-art methods: PROTEUS2 [14], RaptorX [16], Jpred [12], and PSSP-MVIRT [19].…”

“…To evaluate the effectiveness of our model, we used the same benchmark dataset commonly used as a “gold standard” dataset in several studies [19, 28]. The dataset contains 5,772 secondary structures of protein data with three structural states: Helix (H), Strand (E), and Coil (C).…”

“…residues long [19,38,39] . To investigate if our model had length biases for peptide secondary structure prediction, we further explored whether peptide length affected the performance of our model.…”

“…Singh et al [1] proposed PEP2D, the first peptide-specific secondary structure predictor that trains a random forest (RF) model with peptide sequential and evolutionary data and achieves good performance. Recently, Cao et al [19] designed PSSP-MVIRT (Peptide Secondary Structure Prediction based on Multi-View Information, Restriction and Transfer learning) for the prediction of peptide secondary structures, employing CNNs and BIGRU to learn high-latent features and introducing transfer learning to overcome the lack of training data. In addition to the aforementioned methods, there are several other peptide structure prediction methods, such as PEP-FOLD [20].…”