Prediction of Protein Hotspots from Whole Protein Sequences by a Random Projection Ensemble System

Abstract

Hotspot residues are important in the determination of protein-protein interactions, and they always perform specific functions in biological processes. The determination of hotspot residues is by the commonly-used method of alanine scanning mutagenesis experiments, which is always costly and time consuming. To address this issue, computational methods have been developed. Most of them are structure based, i.e., using the information of solved protein structures. However, the number of solved protein structures is extremely less than that of sequences. Moreover, almost all of the predictors identified hotspots from the interfaces of protein complexes, seldom from the whole protein sequences. Therefore, determining hotspots from whole protein sequences by sequence information alone is urgent. To address the issue of hotspot predictions from the whole sequences of proteins, we proposed an ensemble system with random projections using statistical physicochemical properties of amino acids. First, an encoding scheme involving sequence profiles of residues and physicochemical properties from the AAindex1 dataset is developed. Then, the random projection technique was adopted to project the encoding instances into a reduced space. Then, several better random projections were obtained by training an IBk classifier based on the training dataset, which were thus applied to the test dataset. The ensemble of random projection classifiers is therefore obtained. Experimental results showed that although the performance of our method is not good enough for real applications of hotspots, it is very promising in the determination of hotspot residues from whole sequences.

Hot spot identification by means of classification methods employing wavelet transform-based features

“…Robetta [6,10], HotPoint [2,3,5], MAPPIS [1], KFC [11], SpotOn [12], PredHS [13], iPPHOT [14], the method using docking approach [15], HSPred [16], HEP [17]) or physico-chemical properties of their residues (e.g. method using random projection-based classifier [18], MSCA [19], method applying ensemble learning [20], DICFC [21], iFrag [22]). Most of the aforementioned algorithms require knowledge of the protein structure, which is a significant drawback of these methods because the protein structure has been determined only for a limited number of proteins.…”

Machine Learning Approaches for Protein–Protein Interaction Hot Spot Prediction: Progress and Comparative Assessment

“…Physicochemical features (e.g. hydrophobicity, hydrophilicity, polarity and average accessible surface area) from the AAindex1 database [ 14 ] are extracted to predict hot spots [ 15 , 16 ]. Position-specific scoring matrices (PSSMs) are a commonly used sequence feature that can be obtained from NCBI non-redundant databases via PSI-BLAST [ 17 ].…”

“…Hu et al [ 58 ] proposed a protein sequence-based model, in which the classifier is implemented by the improved IBK (Instance-based k means) algorithm of the k-nearest neighbors, which overcomes the shortcomings of the recent neighbor algorithm, which is sensitive to some data. Jiang et al [ 16 ] also proposed a sequence-based model, using the IBK algorithm to obtain a better random projection set through the training set.…”

Special Protein Molecules Computational Identification

“…Most of the works focused on the hotspot predictions from a curated small partial dataset of the whole protein sequences [ 13 ]. In Jiang’s work [ 14 ], the issue of hotspot determination was approached from whole natural protein sequences, and a random projection ensemble system based on k nearest neighbor algorithm to identify hotspot residues by sequence information alone was developed. Experimental results showed that although this method did not perform well enough in the real applications of hotspots, it was very promising in the determination of hotspot residues from whole sequences.…”