A Correlation-Based Solution to Feature Selection in Protein BioInformation

Feature selection techniques have become an apparent need in many bioinformatics applications. In this article, taking into account the protein-protein interactions in sequence and spatial structure of protein, we propose a new correlation based method to capture these correlations between features to check feature redundancy, and implement a corresponding feature selection algorithm CBPFS. Features selected via our method provide a better efficiently, achieving high degree of dimensionality reduction and preserving the original semantics of the bioinformatics characters, hence, offering the advantage of interpretability by a domain expert and improving better accuracy.