关键词: GPCR跨膜螺旋，序列相似性，形变建模，连续概率分布

Abstract: Transmembrane helix is main feature of the GPCR, and the accuracy of single helix prediction directly affects the prediction is of entire GPCR structure. It is a challenge problem to predict the distortion of the GPCR helix The distortion is represented by using the kink residue place and the angle of two fragment helix around the distortion position.Based on the all known GPCR's helixes structure currently, the helixes are clustered according to the helix sequence similarity, and the bend angles in each cluster are modeled by using continuous von mises probability distribution. The modeled GPCR hM kink angles arc tested by using the regression and forecast testing method. Based on this article's model,only fifteen times sampling would have a sample result close to the native TM distortion angle,which will help to improve the hM-helix structure prediction.

Key words: GPCR transmembranc helix, Sequence similarity, Distortion modeling, Continuous probability distribution