A Feature Extraction Method in Large Scale Prediction of Human Protein-Protein Interactions using Physicochemical Properties into Bi-gram

Since the complete sequencing of the human genome, there is a huge volume of biological datas. This is a solution to increase our knowledge on the functioning of our organism. One of the important biological processes to elucidate is protein-protein interaction (PPI), which is at the heart of the entire interaction system between the different biochemical molecules in the cell. Therefore, several feature extraction techniques have been proposed in the past to accurately extract relevant information from protein sequences in large volumes of datas. However, these extraction techniques are sometimes unusable, and the relevance of the extracted information is limited. In this study we propose a feature extraction method in large volumes of data based on bigram directly obtained from two physicochemical properties of amino acids that are hydrophobicity and hydrophilicity. The results of this research were evaluated from commonly used databases and show an improvement of more than 7% on existing methods.