Abstract:Abstract Protein fold recognition is a critical step toward protein structure and function prediction, aiming at providing the most likely fold type of the query protein. In recent years, the development of deep learning (DL) technique has led to massive advances in this important field, and accordingly, the sensitivity of protein fold recognition has been dramatically improved. Most DL-based methods take an intermediate bottleneck layer as the feature representation of proteins with new fold types. However, this strategy is indirect, inefficient and conditional on the hypothesis that the bottleneck layer’s representation is assumed as a good representation of proteins with new fold types. To address the above problem, in this work, we develop a new computational framework by combining triplet network and ensemble DL. We first train a DL-based model, termed FoldNet, which employs triplet loss to train the deep convolutional network. FoldNet directly optimizes the protein fold embedding itself, making the proteins with the same fold types be closer to each other than those with different fold types in the new protein embedding space. Subsequently, using the trained FoldNet, we implement a new residue–residue contact-assisted predictor, termed FoldTR, which improves protein fold recognition. Furthermore, we propose a new ensemble DL method, termed FSD_XGBoost, which combines protein fold embedding with the other two discriminative fold-specific features extracted by two DL-based methods SSAfold and DeepFR. The Top 1 sensitivity of FSD_XGBoost increases to 74.8% at the fold level, which is ~9% higher than that of the state-of-the-art method. Together, the results suggest that fold-specific features extracted by different DL methods complement with each other, and their combination can further improve fold recognition at the fold level. The implemented web server of FoldTR and benchmark datasets are publicly available at http://csbio.njust.edu.cn/bioinf/foldtr/.

Recent Progress in Machine Learning-Based Methods for Protein Fold Recognition

A short review of protein fold recognition methods

Protein Fold Recognition based on Multi-view Modeling.

MLDH-Fold: Protein Fold Recognition Based on Multi-View Low-Rank Modeling

Distance-based protein folding powered by deep learning

Recent Progress of Protein Tertiary Structure Prediction

Protein Fold Pattern Recognition Based on Ensemble Classifiers

Protein Structure Prediction: Challenges, Advances, and the Shift of Research Paradigms

Recent Advances and Challenges in Protein Structure Prediction

A Review of Protein Function Prediction Under Machine Learning Perspective

Advancing the Accuracy of Protein Fold Recognition by Utilizing Profiles from Hidden Markov Models.

Machine learning for protein folding and dynamics

Protein Fold Recognition Based on Sparse Representation Based Classification.

Improving protein fold recognition using triplet network and ensemble deep learning

Exploring the computational methods for protein-ligand binding site prediction

Improved Method for Predicting Protein Fold Patterns with Ensemble Classifiers.

A Survey on Algorithms for Protein Contact Prediction

A Review of Protein Structure Prediction using Deep Learning

Enhanced Protein Fold Prediction Method Through a Novel Feature Extraction Technique.

Protein folding in the modern era: a pedestrian's guide

Advances in Exploration of Machine Learning Methods for Predicting Functional Class and Interaction Profiles of Proteins and Peptides Irrespective of Sequence Homology