Bioinformatics Of Potassium Channels

Potassium channels (K+ channels) are the most diverse group of the ion channel family and are important targets for drug design. They play critical roles in cellular signaling processes and are related to many diseases. So far many biologists have focused their research on studying the relationship between structures and functions of K+ channels. Due to the personality of K+ channels and the experimental limitations, it is difficult to crystallize K+ channel proteins and obtain high-resolution structural data for K+ channel proteins. Therefore, structure-from-sequence predictive methods are demanded to get the information of secondary or tertiary structures and functions of K+ channels. In this paper, we analyzed the signature sequence in P-domain of K+ channels and the similarities of families and subfamilies of K+ channels by multiple sequence alignment. The results indicate that it is difficult to classify K+ channels based on signature sequence in P-domain or the similarities of K+ channels. So We put forward a computational tool'PreK-ClassK-ClassKv', which is a dipeptide composition based support vector machine (SVM) method, to judge whether a novel protein is a K+ channel protein (PreK) and which family (ClassK) or subfamily (ClassKv) it belongs to if it is. The structure and functions of a novel protein can be obtained from the databases with knowledge of K+ channels. This method was evaluated by five evaluating indices (total accuracy, sensitivity, specificity, reliability, Matthews's correlation coefficient) using cross-validation test and independent dataset test. The results show that this...