| Summary: | 碩士 === 逢甲大學 === 生醫資訊暨生醫工程碩士學位學程 === 99 === Proteins or the part of regions of proteins which lack stable and indefinable three dimensional structures are always called disordered proteins or disordered regions. Disordered proteins play important roles in many vital cells and frequently involve in key biological processes, such as signal transduction and regulation. However, due to the amounts of disordered regions are rare in proteins and disordered proteins have no stable structure that lead to difficult in prediction. In this study, we present a novel method which is applied support vector machine algorithm (SVM) to predict the disordered regions by using evolutionary information and structure-based profiles. The evolutionary information is from the Position Specific Scoring Matrix (PSSM) which is generated by PSI-BLAST. The structure-based profiles are the scoring matrices of the statistical probability from protein structure-related properties, such as secondary structure, solvent accessible area or three dimensional conformations, etc. 723 non-redundant protein chains which contain disordered regions are used for training and testing by 5-fold cross validation. Finally, the accuracy can be achieved 94% in the prediction of disordered regions in proteins.
|
|---|
