Enhancing Protein Fold Prediction Accuracy Using Evolutionary and Structural Features

Abstract

Protein fold recognition (PFR) is considered as an important step towards the protein structure prediction problem. It also provides crucial information about the functionality of the proteins. Despite all the efforts that have been made during the past two decades, finding an accurate and fast computational approach to solve PFR still remains a challenging problem for bioinformatics and computational biology. It has been shown that extracting features which contain significant local and global discriminatory information plays a key role in addressing this problem. In this study, we propose the concept of segmented-based feature extraction technique to provide local evolutionary information embedded in Position Specific Scoring Matrix (PSSM) and structural information embedded in the predicted secondary structure of proteins using SPINE-X. We also employ the concept of occurrence feature to extract global discriminatory information from PSSM and SPINE-X. By applying a Support Vector Machine (SVM) to our extracted features, we enhance the protein fold prediction accuracy to 7.4% over the best results reported in the literature.

A Segmentation-Based Method to Extract Structural and Evolutionary Features for Protein Fold Recognition

“…It provides the substitution probability of a given amino acid based on its position along a protein sequence. Extracted features from PSSM have been widely used for the PFR and attained promising results [27], [28], [29].…”

“…Features that capture significant global and local discriminatory information and the classification techniques that perform consistently with these extracted features have been used in the literature [2]. A wide range of classification techniques such as, artificial neural networks (ANN) [3], [4], [5], [6], [7], [8], meta classifiers [9], [10], [11], [12], [13], K-nearest neighbors [14], [15], [16], [17], [18] and support vector machines (SVM) [19], [20], [21], [22], [23], [24], [25], [26], [27] have been used for the PFR. Among the classifiers employed to tackle the PFR, using support vector machine have attained the best results [26], [27], [28], [29], [30], [31], [32].…”

“…A wide range of classification techniques such as, artificial neural networks (ANN) [3], [4], [5], [6], [7], [8], meta classifiers [9], [10], [11], [12], [13], K-nearest neighbors [14], [15], [16], [17], [18] and support vector machines (SVM) [19], [20], [21], [22], [23], [24], [25], [26], [27] have been used for the PFR. Among the classifiers employed to tackle the PFR, using support vector machine have attained the best results [26], [27], [28], [29], [30], [31], [32]. Similarly, a wide range of features have been extracted and used to tackle the PFR such as, physicochemical-based features [19], [23], [33], [34], sequence-based features [6], [14], [15], [32] evolutionarybased features [18], [25], [28], [30], and structural-based features [17], …”

“…Similarly, a wide range of features have been extracted and used to tackle the PFR such as, physicochemical-based features [19], [23], [33], [34], sequence-based features [6], [14], [15], [32] evolutionarybased features [18], [25], [28], [30], and structural-based features [17], [18], [23], [35], [36], [37], [38]. Achieved results have shown that the most significant enhancement for the protein fold prediction accuracy has been achieved by relying on the feature extraction approaches rather than the classification techniques being used [4], [15], [19], [27], [28], [29], [39]. In most of the studies that addressed the PFR by feature extraction techniques, global discriminatory information has been represented using the composition of the amino acids feature group (the percentage of the occurrence of the amino acids along the protein sequence divided by the length of protein sequence [19], [22], [30]).…”

A Tri-Gram Based Feature Extraction Technique Using Linear Probabilities of Position Specific Scoring Matrix for Protein Fold Recognition

“…It is also shown that by fusion of features the recognition rates can be improved [30]- [33]. For the latter task case, several classifiers have been developed or used including linear discriminant analysis [34], [35], Bayesian classifiers [2], Bayesian decision rule [36], k-nearest neighbor [25], [37], Hidden Markov model [38], [39], artificial neural network [40], [41], support vector machine (SVM) [6], [20], [21], [42], [43], and ensemble classifiers [20], [33], [41], [44], [45]. Among these classifiers, SVM (or SVM-based for ensemble strategy) classifier exhibits quite promising results [21], [26], [27].…”