In large Volunteer computing for molecular docking projects such as FightAIDS@home, do they use the original crystal structure of ligands in docking or unconfirmed structures of ligand?


I don't know that project's details, so cannot say officially, but undoubtedly it will be

  • crystal structures of protein or models of point mutants based on structures of protein.
  • Computer structure of the ligands based on force-field calculation

Protein structures

Firstly, HIV virus has been extensively studied by structural biologists, so most components have structures.

Secondly, docking on models is generally done only for threaded models based on closely related templates, say a human model of a mouse structure or a recently diverged pair (i.e. conserved proteins that split with the genome double duplication in bony fish). Whereas a given virus protein will likely have poor homology to a known structure and not be a good candidate.

A corner case is core point mutants... If you have a mutation in a protein it is likely going to be on the surface and not core: a model of a point mutant is likely very faithful to an empirical version —and generally you'd want to design drugs to a region that is not variable anyway as you'd want it to work universally. If you have a change in the active site to confer resistance say (i.e. in the core), there is generally a fitness loss to the organism (say lowered catalysis or thermostability) which needs to be compensated with other mutations (epistasis). As this is rarer and very interesting scientifically and medically, these resistant variants will most likely have been crystallised as the conditions for crystallization will most likely be the same.

Small molecule structures

Force field calculations for small molecules are most often valid (there computer Vs empirical RMSD is minute). The partial charges may be off, but that's not something crystals give anyway: In fact a ligand used in docking is (generally/ought to be) energy minimised beforehand. For docking getting charges wrong is the killer.

Crystallization of small molecules is used as validation of a synthesis route, so some may have structures, but they will not be used in a pipeline as a SMILES string is all you need.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.