[apbs-users] Setting up for an asynchronous APBS calculation
Grant Kemp
Hi everyone,
I am having a bit of a hard time getting an APBS calculation up and running. I am working on calculation a highly resolved electrostatic surface of the tunnel of the ribosome. I am hoping to do the calculations just on my laptop if I can before reaching
out the to the computing facilities for some time. I have an i5 (4 core) machine running 64-bit linux with 8Gb of RAM. Of course the calculation on the whole ribosome fails due to insufficient RAM, but when I focus the calculation on the tunnel (by selecting
only the zone within 20A) I can manage calculations with 150mM NaCl. This looks quite nice, but I also wanted to try the calculation in the presence of Mg (preferably 10mM, but I would be willing to do 2-3mM) and I cannot get the calculation to run (I always
get the insufficient RAM message). Eventually, to make sure that looking at only the tunnel is an accurate way to examine our question, I would like to do calculations on the whole ribosome. This seemed like a problem that could be solved by setting up an
asynchronous parallel APBS calculation, but I am struggling with exactly how to do this. The steps I have followed are:
Prepare my pdb using pdb2pqr (no problems)
Use the psize.py script to determine the pdime that I should use
Use the inputgen.py script with the --split option on the input file generated by pdb2pqr into a multiple files for asynchronous APBS calculation
The input file I provide to inputgen.py is attached. When I try to run APBS on one of the generated files in the command line (apbs ribo_para-PE0.in) I get the same insufficient RAM message as I got before. I think I am missing something. Either I need a
script to coordinate the calculation of the PE files together or something goes wrong with the inputgen.py (also attached in case it is wrong) and my generated input files are wrong (they are the same as the unsplit input file except they have the async {0...n}
line added). I hope that I am just missing something simple and that i can be easily explained :)
Thanks in advance!
Grant
Star, Keith
Grant Kemp
Hi Keith,
Sure. I can't actually make it work using any pqr (is this the right file). Maybe this is the problem. I've attached both the pdb and the pqr for you to test. Here are the commands that I have been running from within the directory containing all the files
(i.e. I have the psize and inputgen scripts in the working directory not in the python modules directory):
python ./psize.py tunnel_empty.pqr (I use the "Proc. grid" output from psize as the pdime value in the input script)
python ./inputgen.py --split tunnel_empty.in
This generates a series of input files with the -PE"n" appended to the end. The only difference I can see is that inputgen has added the async "n" line to the input file. To test it I then run:
apbs tunnel_empty-PE0.in
I am guessing that I am either running the inputgen script wrong or I need to use another script to bundle the newly created asynchronous input files. I hope this helps you help me. I really appreciate it (again!).
Best,
Grant
Sent: 20 September 2016 20:39
To: Grant Kemp; apbs-...@lists.sourceforge.net
Subject: Re: [apbs-users] Setting up for an asynchronous APBS calculation