Prediction Of Protein Crystallization Based On Multi - Angle Feature Fusion And Random Forest

Protein is an important material foundation of organism, and it’s also a basic organic of cells. The phenomenon of life is mainly realized through the structure and function of protein. The percentage of protein weight is about 16%～20% of the whole body weight. Each part and each cell of the body has a protein in it. So the protein plays a very important role in the process of life. It’s very important for us to grasp the various attributes of proteins so that we can comprehend the functions of protein, understand various kinds of biochemical reactions, gene expression within n the organisms and drug development.There are three main methods for three-dimensional structure analysis of proteins:X-ray crystallography, Nuclear magnetic resonance, Electron microscopy 3D reconstruction. Among them, the most widely used is the X-ray crystallography.As one of the most popular experimental approaches, the X-ray crystallography method has been utilized to obtain 80～90% of the deposited protein structures in the Protein Data Bank (PDB). However, not all the proteins used for determining structures are crystallizable, which will lead to a low success rate of crystallization projects and lots of time and resources on those non-crystallizable protein were wasted. Hence, developing accurate and effective methods for predicting whether a protein will crystallize is of significant importance. In this study, we propose a new protein crystallization prediction method, among which features derived from physicochemical properties of residues, pseudo amino acids composition (PseAAC) and pseudo position specific scoring matrix (PsePSSM) are combined to form discriminative feature of a protein; random forest is taken as classifier for performing protein crystallization prediction. Experimental results on benchmark dataset over cross-validation test and independent validation test show that the proposed method is comparable to the state-of-the-art protein crystallization predictors, which demonstrates the efficacy of the proposed method.