A Scale of p-Preferences for Structure-Activity Predictions in Membrane Proteins
Davor Juretić, Roko Pešić · University of Zagreb University Computing Centre (SRCE) · 1995
Physical chemists are often interested in predicting molecular properties from the known molecular structure.In this paper, a different approach is taken to predicting secondary structure features, membrane folding motifs and functionally important amino acid residues in membrane proteins, starting from physical, chemical 01' statistical properties of twenty natural amino acid types.Empirical knowledge of protein structure and property scales is combined through the preference function method so as to predict the secondary structure of membrane proteins in the three state model (u-helix, B-sheet,coil 01' turn).Of the 140 property scales examined, our own scale of ~-sheet preferences, extracted from porins and defensins, is the best in predicting the sequence location of transmembrane helices.It is also shown that functionally important amino acids located in such helices can be predicted with satisfactory accuracy for the case of bacteriorhodopsin and the potassium channel subunit, where identity of amino acid residues involved in the ion transport activity is fairly well established.