Prediction of protein folding kinetic states using fuzzy back propagation method

Abstract

Protein folding process is extremely vital in major the molecular function. The kinetic order of protein folding chooses whether the molecule reaches its native structure through intermediates or not. They can either fold without stable intermediates (2State/2S) and with stable intermediates (3State/3S). This is generally determined using equilibrium denaturation research and is often time consuming and tedious. Moreover, the unfolding appliance of large number of Proteins available in the PDB are found unknown. Therefore, it created interest and directed us to predict and classify the folding mechanism as two state or three state (Multiple State). We developed the classification models using Fuzzy Back Propagation Network (FBPN) with the known attributes (Protein length (PL), hydrophobicity, hydrophilicity, secondary structural components). The models performed fairly well for predicting two state and three state folding using the well-known variables. The FBPN model produced accuracy of 83% for cross validation. This method thus can intensely assist as a outline for predicted monomer and dimer structures with unknown folding appliance for further confirmation through investigational studies.