Self-concerted hydrophobicity scale based on the ACDlogP of a combination of 8000 tripeptides and its application for identification of protein active sites

Partition coefficients, expressed as logP, were calculated using Advanced Chemistry Development software (ACDlogP) [1] of all combinations of three amino acids (8000 tripeptides). Using our proprietary script, we generated a combination of 8000 tripeptides of 20 amino acids in the FASTA format and, subsequently, full atom Cartesian coordinates were generated. The ACDlogP’s of the generated tripeptides were calculated. Using the Kyte-Doolittle amino acid hydrophobicity scale [2, 3], the value of the correlation coefficients with the calculated ACDlogP values was determined. Hydrophobicity values were assumed as the arithmetic mean of the hydrophobicity of the three amino acids in the tripeptide. Optimisation of the theoretical hydrophobicity scale by minimisation of the correlation coefficient between the calculated ACDlogP values and the hydrophobicity for the tripeptides provided amino acid hydrophobicity; on this basis, the amino acids were divided into 7 groups. The new scale was normalised and implemented using the fuzzy-oil-drop model method to determine the theoretical protein active site of the 1HCK protein based on lipophilic hot spots on the protein surface. The results were compared with the respective results for the Kyte-Doolittle scale and the actual active site with ATP as the ligand.