[gmx-users] Re:Ka/Kd

[Justin Lemkul]

On 11/11/12 7:47 AM, lloyd riggs wrote:
> Dear Justin,
>
> Thanks, Just a normal delta G then.  I wondered if it also was doing a matrices from 2 columns if you supply something like just the 2d output from covar/anaeig (PCA).  I did find a simple python 6 liner for just doing PCA with the first (or any 2) derivatives just using numpy, and some module written by tsjerk Wassenaar posted a few years ago (a Gromacs plugin called g_entropy) which all together seem to meet my needs.  Also (the above person) has a number of small python scripts for Gromacs for matirces, PCA, and just turning .xpm data from g_sham into a number matrices to put into a plotting tool (such as Scidavis or something).  There on a web page, and a few dozen just scattered in older gromacs postings.
>
> If I had a request for Gromacs dev though it would be to multithread the vector portions of the analytical tools.  A single Intel i7 or AMD only uses one of the 8 cores, so it takes several hours, for each component times each of however trajectories are looked at.  Most everything else is a minute or two only using a single core.
>

Framework for parallelizing analysis tools is in the works, with a target of 
version 5.0 for implementation, since the code is undergoing a major overhaul 
for that release, anyway.  It will be a long time before that release, though, 
but it's in the works.

-Justin

-- 
========================================

Justin A. Lemkul, Ph.D.
Research Scientist
Department of Biochemistry
Virginia Tech
Blacksburg, VA
jalemkul[at]vt.edu | (540) 231-9080
http://www.bevanlab.biochem.vt.edu/Pages/Personal/justin

========================================