:
In this letter the newly introduced approach based on the radial distribution function (RDF) weighted by the number of va-lence shell electrons is applied for a series of HIV-1 protease enzyme and its complexes with inhibitors to evaluate the influ-ence of hydrogen atoms on the performance of the model. The multiple linear regression method was used for the selection of the relevant descriptors. Two groups of residues having dominant contribution to the RDF descriptor are identified as relevant for the inhibition. In the first group are residues like Arg8, Asp25, Thr26, Gly27 and Asp29, which establish direct interaction with the inhibitor, while the second group consists of the amino acids at the interface of the two homodimer sub-units or with the solvent. The crucial motif pointed out by our approach as the most important for inhibition of the enzyme’s activity and present in all inhibitors is hydroxyl group that establish hydrogen bond with Asp25 side chain. Additionally, the comparison to the model without hydrogen showed that both models are of similar quality, but the downside of the current model is the need for the determination of residues’ protonation states.