New Biological Insights from Better Structure Models.
A novel hydrophobic penalty function of proteins is proposed and assessed with several test cases. The number of residues in a defined sphere around a certain residue is averaged over the data set proteins. Differences between the standard values thus obtained and calculated values are summed up, residue by residue, with the weight of standard deviations to give the penalty value. This penalty function is applied to the structures of randomly shuffled sequences, incorrectly folded structures and partially denatured structures displayed on a graphics terminal, and is shown to discriminate the native structure from others fairly well, although the present parameter set is tuned for proteins of about 100-150 residues. From the results of present study and the known correlation with other hydrophobic parameters of amino acids, the penalty function can be considered as a practical amino acid residue-level hydrophobic penalty function.