Re: Is there really no way of automating pair interaction calculations without invoking NAMD for each pdb file separately?

From: Brian Radak (brian.radak_at_gmail.com)
Date: Fri Nov 02 2018 - 09:28:21 CDT

It might be worth turning off explicit FFT-tuning in this case
("FFTWEstimate yes"), as I suspect this is what is causing the large
overhead. Alternatively, you can make sure that the jobs share the same FFT
"wisdom" file ("FFTWWisdomFile <filename>" or just copy the file to the new
working directory).

However, I think the keyword routes will only work if you are operating
from a script because you need keywords which are probably not part of the
namdenergy plugin. Peter's suggesting of saving the temporary scripts and
then modifying them is a good one.

HTH,
BKR

On Fri, Nov 2, 2018 at 9:32 AM Asghar Razavi <asr2013_at_med.cornell.edu>
wrote:

> Thank you Peter and Steve for your comments. I am following Steve's
> approach in using namdenergy for the calculations. It is still invoking
> NAMD for each set of calculations, but, at least I don't need to create pdb
> files by hand.
>
> @ Peter: I do have all my coordinates in a single dcd file. I don't need
> to update selection of atoms for each step. However, I want to calculate
> interaction energies for > 100 pairs of residues. It would help a lot if
> NAMD allowed reassignments of interacting residues, so that calculations
> could be done faster. My system is large and invoking NAMD to initiate
> calculations every time takes more than 10% of the total calculation time--0000000000009c8cd10579af59d6--

This archive was generated by hypermail 2.1.6 : Fri Dec 06 2019 - 23:20:12 CST