Prediction of Protein Secondary Structure Based on NMR Chemical Shift Data Using Support Vector Machines

Protein secondary structure detection is an intricate problem which depends on several parameters of a polypeptide chain and its environment and has a great effect on the accurate determination of protein functionality in living organisms. Statistical learning approaches have been used to tackle the problem extensively and many considerable results have been achieved, which encourages the researchers to continue exploring the track. Support vector machines are among the interesting tools of machine learning, which have been used in different fields of computational molecular biology. This paper aims to combine the power of SVMs with the informative chemical shift data to distinguish the secondary structure of the proteins. The results show a good accuracy of the approach regarding different structures, especially in detection of turns and sheets.