Detection Of Protein-protein Interaction Based On Rotation Forest

As carrier of life activities, protein plays an essential role in organism. And protein-protein interactions(PPIs) exist in biogenic activities universally, such as metabolism of cell, cell cycle changes, regulation of hormone and enzyme catalysis. Therefore, research in PPIs help people to understand not only the progress of life, but also the discovery in pathogenesis, therapeutic development and searching drug targets. Targeting at the defects of high-throughput experiments in bioscience, combining data mining technology and physicochemical property of protein sequence are proposed to study on detecting protein-protein interactions(PPIs).The main works are as following:1. In this paper, research background and significance of PPIs are introduced. Also, PPIsrelated knowledge and processes of data mining are stated.2. Combining physicochemical property and information of protein sequence, the physicochemical property response matrix is built. Then local phase quantization method is employed to extract the features of protein sequence.3. Combining physicochemical property and information of protein sequence, quasi residue couple method is employed to extract the features of protein sequence.4. Three prevalent datasets are used for validation. Rotation Forest model are proposed as classifier, and the results are compared with the present study results. Also, using the same feature extraction method, random forest model is trained and its results are compared with ones of rotation forest model. Finally, we conclude that combining the feature extraction methods and rotation forest model can yield a promising results.