Martin64: "Crunching faah26205 - so either there was an extension to Experiment 37, or Experiment 38 has started."
Or they left a gap in batch numbers from 26178 - 26199 and started Exp 38 at 26200 or thereabouts.

Observing the finish-up of 1 of the new WUs, I noticed that it counted "Progress" right up to about 99.5%, with "To Complete" down to less than 1m before it ended & uploaded. WUs from previous experiments usually ended with less "Progress" and about 3 - 10m remaining "To Complete".

Hello Rickjb,

You're welcome.

When other scientists have systematically studied this general issue, they discovered that some docking programs are more accurate against particular types of drug targets, while other docking programs are more accurate against other types of drug targets. There is no single docking program that is "more valid" for all the different types of drug targets. This is one of the reasons why I like to examine potential inhibitors with two different docking programs (that is, AutoDock and the new "AutoDock Vina"). This issue is especially important with the GO Fight Against Malaria project, since that project will screen compounds against many very different types of drug targets.

When I have compared AutoDock results to Vina results for a particular target, I generally do observe some overlap between the best ligands that each program predicts. I focus the most on these ligands that score the best with both programs, but I also select and order some compounds that only score well with one program. This is a tricky issue that still needs a lot of exploration, and there probably won't be one approach that will always produce the best predictions against all targets of interest.

Yes, if we only paid attention to the compounds that scored well with both types of docking programs, we would miss some potentially interesting compounds (that is, there would be some "false negative" predictions). But that's why I inspect the best results from each program, as well as focusing on the compounds that scored well with both of them.

I don't have an answer for the question about single-redundancy jobs that use Vina. This is an issue that I need to discuss further with Professor Art Olson and with the IBM team.

Some virtual screens that use AutoDock do produce many false positive results, while other screens don't. But I've generally had good success rates in the virtual screens that I have performed with AutoDock. It depends on many different factors, such as: the particular target that the compounds are screened against, the particular compounds that are being screened against that target, the manner in which the input files for the target and for the compounds were prepared, the run parameters used within the calculations, how the results are sorted to fish out the top-ranked compounds, who is visually inspecting the top results, and how they decide to select the top compounds for subsequent biological assays.

CHARMM is a Molecular Dynamics program--MD simulations are very different than docking calculations, but they can complement each other well. Several labs have shown that MD simulations can be used to help analyze and re-rank the results that are predicted by different docking programs, but these MD simulations are very computationally expensive. At some point in the future, we might use MD simulations (with either AMBER or CHARMM) to study and re-rank the virtual screening results generated for FightAIDS@Home or for GO Fight Against Malaria, but that possibility is far down the road.

Thank you very much for your interest and your support. And have a very happy New Year!

Best wishes,
Dr. Alex L. Perryman