[Computational methods for prediction of structure of membrane proteins using their amino acids sequences].

The structure of membrane proteins is interesting because of their functional properties that are important to medicine and pharmacology. The feature and an organic property of polytopic membrane proteins is the repetition of transmembrane regions consisting of hydrophobic amino acids. The ordered repetition--periodicity--can be identified by the Fourier method, applied to a digital image of symbolic sequence of amino acids in a protein. In this work it was carried out for the 24 transmembrane proteins, for 14 of them successfully. If the repetition of transmembrane regions is ordered insufficiently--non-periodic, then a different method is supposed to use for its detection--the method of multiple (4-5 times) averaging of function of hydrophobicity of the protein within a "window" with width 9-11 aa moved along the sequence. This new method was applied to the same 24 transmembrane proteins (for 19 of them successfully) and it was shown to be more suitable (than the Fourier method) for predicting of the secondary structure of such proteins and functional properties corresponding to it.