Abstract:Background and Objective: Protein structural class prediction is a first and key step in protein structure prediction and has become an active research area in biochemistry and bioinformatics. An important aspect for this prediction task is exploring good feature representation. Prior works have demonstrated the effectiveness of the PSI-BLAST profile based feature extraction methods especially for low-similarity protein sequences. However, the prediction accuracies still remain limited. This highlights the need for keeping on exploring the potential of evolutionary information. Method: In this study, three novel sequence evolutionary modes of pseudo amino acid composition (PseAAC) are proposed and optimized by a two-stage feature selection process based on recursive feature elimination strategy. The selected top-ranking features are then fed into a linear kernel support vector machine classifier to predict the protein structure class. To evaluate the performance of the proposed method, jackknife tests are performed on three widely used low-similarity benchmark datasets (25PDB, 1189 and 640). Results: With comprehensive comparison with the current state-of-the-art methods, the proposed method achieves superior performance. The overall accuracies on 25PDB, 1189 and 640 datasets are 96.2%, 97.9% and 99.5%, which are 1.9%, 1.5% and 2.3% higher than previous best-performing method. Conclusion: The satisfactory prediction accuracies achieved by the proposed method are attributed to the specially designed sequence evolutionary modes of PseAAC and the effective feature selection strategy, which cover more discriminative sequence order information. It is anticipated that our method would be helpful in other prediction problems in protein research.

A novel kernel sequence auto-correlation functions applying to prediction of non-homologous protein structural classes

Prediction of Functional Class of Proteins and Peptides Irrespective of Sequence Homology by Support Vector Machines.

Using Correlation Analysis and Nonnegative Matrix Factorization to Predict Protein Structural Classes via Position-Specific Scoring Matrix

Application of Improved Three-Dimensional Kernel Approach to Prediction of Protein Structural Class.

A Correlation-Coefficient Method to Predicting Protein-Structural Classes from Amino-Acid Compositions

Using Pseudo-Amino Acid Composition and Support Vector Machine to Predict Protein Structural Class.

Protein Structure Class Prediction Based on Autocorrelation Coefficient and PseAAC

An Ensemble Classifier of Support Vector Machines Used to Predict Protein Structural Classes by Fusing Auto Covariance and Pseudo-Amino Acid Composition

Prediction of Protein Structural Class Based on Different Autocorrelation Descriptors of Position-Specific Scoring Matrix

On Using Physico-Chemical Properties of Amino Acids in String Kernels for Protein Classification Via Support Vector Machines.

Prediction of Protein Structural Classes Based on Correlations of Amino Acid Residues

A Novel Conjoint Triad Auto Covariance (CTAC) Coding Method for Predicting Protein-Protein Interaction Based on Amino Acid Sequence

A Novel Predictor for Protein Structural Class Based on Integrated Information of the Secondary Structure Sequence.

Predicting Protein Structural Class with Pseudo-Amino Acid Composition and Support Vector Machine Fusion Network.

Predicting Protein Structural Class for Low-Similarity Sequences via Novel Evolutionary Modes of PseAAC and Recursive Feature Elimination

Prediction of protein structural class using novel evolutionary collocation-based sequence representation.

Prediction of Protein Structural Classes Based on one DFS Algorithm

Predicting Protein Structural Classes for Low-Similarity Sequences by Evaluating Different Features

Protein remote homology detection by combining Chou’s distance-pair pseudo amino acid composition and principal component analysis

Prediction of Protein Structural Class Using PSI-BLAST Profile Based Collocation of Amino Acid Pairs

Protein-Protein Interactions Prediction Using A Novel Local Conjoint Triad Descriptor Of Amino Acid Sequences