A kernel model with conditional moving windows for the prediction of transmembrane helices in proteins

The determination of some proteins structure at high resolution often results difficult from an experimental point of view. It is true especially for many integral membrane proteins and intrinsically disordered proteins. In this context, we try to develop a statistical technique alternative to the classical methods used to calculate hydropathy graphics, called Kernel Windows Method with Conditional Moving Windows (KWMCMW); the latter has been implemented in order to improve the degree of accuracy of the transmembrane helices (TMH) prediction. The KWMCMW introduces the effect of distance in calculating the hydropathy value for each amino acid (AA) residue in the sequence. Consequently, this new method is extended to a peripheral membrane protein with partially unknown structure. The implementation of specific software written in Python language makes the method suitable for application to any membrane protein.