3D protein structure prediction using Imperialist Competitive algorithm and half sphere exposure prediction

• ICA algorithm, HSEpred, and PSIpred were utilized for protein folding problem.
• A new method for revolutionary changes was designed to check more important states.
• The results showed improvement in the accuracy of the predicted structures.
• Segments structures’ effects on the potential energy is larger than directs’.

Predicting the native structure of proteins based on half-sphere exposure and contact numbers has been studied deeply within recent years. Online predictors of these vectors and secondary structures of amino acids sequences have made it possible to design a function for the folding process. By choosing variant structures and directs for each secondary structure, a random conformation can be generated, and a potential function can then be assigned. Minimizing the potential function utilizing meta-heuristic algorithms is the final step of finding the native structure of a given amino acid sequence. In this work, Imperialist Competitive algorithm was used in order to accelerate the process of minimization. Moreover, we applied an adaptive procedure to apply revolutionary changes. Finally, we considered a more accurate tool for prediction of secondary structure. The results of the computational experiments on standard benchmark show the superiority of the new algorithm over the previous methods with similar potential function.