ADL: Question

David M. Toth david.toth at centre.edu
Mon Sep 28 04:59:50 PDT 2015


Hi Farzaneh.  I'm a computer scientist who works with Vina in conjunction with biologists and chemists on drug discovery.  I have spent some time looking through the Vina code over the last few years and here's my take:

1 - If it does not get ported to run on GPUs (see number 2), then it's as parallel as it needs to be already.  If you have a multicore CPU in your computer with 8 cores, launch 8 copies of Vina simultaneously, each testing a different molecule against your target, and that's the most efficient way to do it.  That will give you linear speedup, which is the holy grail in parallel computing.  You can use xjobs on Linux to constantly keep each CPU core running a copy of Vina so they are never idle.  This is what I do on my 64-core servers that I use for drug discovery work.  If you have access to a supercomputer or a big cluster, you can do a similar thing.  There are also a couple versions of Vina out there (I think VinaLC and some other one I can't remember the name of) that are ports of Vina to run more easily on a supercomputer.

2 - I looked into porting Vina to GPUs with CUDA about 4 years ago and quickly concluded that it would (1) take somebody who really knew the Vina code inside and out to do it without horribly breaking it (in other words, not me) and (2) I wasn't convinced that a GPU version would run well given the way GPUs operate.  I'm concerned that we'd actually get slowdown rather than speedup.


Hope that helps.

Dave Toth


--------------------------------------------------
David Toth
Assistant Professor of Computer Science
Centre College
david.toth at centre.edu
1-859-238-5405
--------------------------------------------------




-----Original Message-----
From: autodock-bounces at scripps.edu [mailto:autodock-bounces at scripps.edu] On Behalf Of farzaneh mohamadyar
Sent: Monday, September 28, 2015 7:47 AM
To: autodock at scripps.edu
Subject: ADL: Question

Dear members,

I am a biologist working on drug design. I've just started to study the source programs of Auto Dock Vina، as I would like to use it in a research.
I'm writing to you because I wonder if it is possible to parallelize the software? I mean can I change the source codes such that I could do n number of docking for n number of drugs at the same time?

I appreciate your consideration and look forward to hearing from you.
Sincerely
________
________________________________________________
--- ADL: AutoDock List  --- http://autodock.scripps.edu/mailing_list ---



More information about the autodock mailing list