Prediction of protein long-range contacts using GaMC approach with sequence profile centers

In this paper, we apply an evolutionary optimization classifier, referred to as genetic algorithm-based multiple classifier (GaMC), to the long-range contacts prediction. As a result, about 44.1% contacts between long-range residues (with a sequence separation of at least 24 amino acids) are founded around the sequence profile (SP) centre when evaluating the top L/5 (L is the sequence length of protein) classified contacts if the SP centers are known. Meanwhile, with the knowledge of sequence profile center and the GaMC method, about 20.42% long-range contacts are correctly predicted. Results showed that SP center may be a sound pathway to predict contact map in protein structures. Availability- http://mail.ustc.edu.cn/~bigeagle/gamc.htm.