Closed toby-aqemia closed 6 months ago
Vina uses its own types internally, and carbons that are bound to some heteroatoms are not considered hydrophobic. So, the system with four atoms has four hydrophobic carbons, but in the original molecule atom 18 (atom 3 in the four-atom system) is not hydrophobic because it is bonded to nitrogen.
Ah I missed that, I thought it was all done upstream of the pdbqt file. Thank you very much for your help!
I am experiencing some unexpected behaviour with the Vina scoring function, in particular the hydrophobic interaction term. From what I understand the score is calculated as the following sum: $$c=\sum{i < j} f{t_i tj}(r{ij})$$ And the hydrophobic term is always non-negative (but then weighted to become negative). For this reason I expect that adding atoms should always increase (or leave unchanged) the hydrophobic interaction term. However this is not what I observe. With the following example I find the interaction term for a molecule to be
1.452
while the interaction term for a subset of atoms is1.766
. Note that I extract the subset by deleting lines in the pdbqt file so the atom types don't change (although the molecule becomes unrealistic). I thought I could be observing some difference due to intra-molecule interactions but from what I understand these aren't included in thelig_inter
part of the score output.Could it be to do with how I've built my toy example with unrealistic molecules and proteins? Or have I fundamentally misunderstood some aspect of how the scoring function is built? I would much appreciate any help to see what I am missing.
Code to reproduce
protein_fragment.pdbqt
mol.pdbqt
subset_of_mol.pdbqt made up of atoms 4, 5, 18, and 19 of the molecule above.