Improving protein fold recognition using triplet network and ensemble deep learning

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/.

An Analysis of Protein Language Model Embeddings for Fold Prediction

“…In addition, in order to compare with DeepFRpro [38], CNN-BGRU-RF+ [44], FoldTRpro [43], FSD_XGBoostpro [43], and FoldHSphere [45], which apply a random forest (RF) ensemble, we implemented and tested the same RF strategy in our ensemble approach (see Materials and Methods section). It must be noted that these results cannot be directly compared to the previous ones, as this approach involves additional training of the RF models on the test set in a 10-stage cross-testing manner.…”

“…Finally, we compare the performance of our best individual models, as well as the ensemble strategy we propose, against the state-of-the-art results for fold recognition and fold classification. First, we compare to several methods intended for the PFR task, which can be grouped into three categories: (i) alignment and threading methods such as PSI-BLAST [104], HHpred [20], RAPTOR [23], BoostThreader [22], SPARKS-X [24], MRFalign [21], and CEthreader [28]; (ii) machine learning methods such as FOLDpro [29], RF-Fold [31], DN-Fold [31], RFDN-Fold [31]; and (iii) deep learning methods such as DeepFR [38], CNN-BGRU [44] VGGfold [42], FoldTR [43], and FoldHSphere [45]. Table 4 shows the PFR accuracy results achieved by these methods on the LINDAHL test set, as well as the best performing model ProtT5 + LAT and the average ensemble.…”

“…In addition, in order to compare with DeepFRpro [38], CNN-BGRU-RF+ [44], FoldTRpro [43], FSD_XGBoostpro The top 1 and top 5 accuracy (%) results are provided for the full test set, and the family, superfamily and fold subsets. Boldface indicates best performance per set.…”

MoRF-FUNCpred: Molecular Recognition Feature Function Prediction Based on Multi-Label Learning and Ensemble Learning

“…Ensemble learning is used in many protein tasks and has good performance, such as recognition of multiple lysine PTM sites and the different types of these sites ( Qiu et al, 2016a ), recognition of phosphorylation sites in proteins ( Qiu et al, 2016b ) and recognition of protein folds ( Liu et al, 2021 ). The ensemble model usually has better performance than individual predictors.…”

Learning Protein Embedding to Improve Protein Fold Recognition Using Deep Metric Learning

“…In recent years, many protein fold recognition methods have been developed. These methods can be generally classified into two major categories: , (1) template alignment methods, which calculate the similarity between query protein and template protein based on sequence–sequence similarity or sequence–structure similarity; and (2) machine learning methods, which treat the matching of the query protein and template protein as a binary classification problem. Regardless of the strategy used, extracting representative features from protein sequences is the key to improving the accuracy of protein fold identification.…”