SECONDARY STRUCTURE OF MEMBRANE-PROTEINS - PREDICTION WITH CONFORMATIONAL PREFERENCE FUNCTIONS OF SOLUBLE-PROTEINS

Conformational preference functions are derived from the statistical analysis or the data base of soluble protein structures. These functions use local sequence information to modify the Chou-Fasman's reference of a given residue in a protein for secondary conformation. The secondary structure prediction algorithm that compares preferences is constructed. For the testing set of 14 membrane polypeptides the prediction accuracy is 78% in the three state model and 90% for the alpha-helix residues alone. Correlation coefficients are 0.58 and 0.57 for the alpha-helix and turn structures, respectively.