Question: How to know a program I use is parallelizable?
0
gravatar for mparadis92
3.8 years ago by
mparadis9240
Canada
mparadis9240 wrote:

I'm an intern in a bioinformatic lab for the summer and I just got permission to use a super-computer to do protein disorder jobs. I know how to use this program on my local computer with linux and it's pretty simple. I downloaded it, make install, run the script with the right PATHs to the files I want to use et voilà.

My question is quite simple but since I'm new to all of this, I don't really know where to look : How can I know if I can use this program in parallel? I ask this because apparently I need to know that before I can submit a job. On the supercomputer there are nods with 8 cores. According to my colleagues, knowing if my job is parallelizable is very important since it will determine whether I use one core or all 8, which will affect time and resources.

I hope this makes sense and I'm sorry if it's a dumb question but I'm kind of lost here... Thanks for the help! :)

(the program I use is DISOPRED3, available for free)

ADD COMMENTlink modified 3.8 years ago by Nicola Casiraghi440 • written 3.8 years ago by mparadis9240
1

The software man page generally has information on how to best parallelize runs. Quite a few use OpenMPI and such, so you could read through their documentation. If the developers have a github repo, you can read through the README there and look at issues. You could also email them.

ADD REPLYlink written 3.8 years ago by RamRS20k

Sadly there is not a lot of documentation. I searched a lot and couldn't find a definitive answer. I thought maybe there were ways to test this or something. Thanks for the answer!

ADD REPLYlink written 3.8 years ago by mparadis9240
1

cross-posted: http://stackoverflow.com/questions/30645791

ADD REPLYlink written 3.8 years ago by Pierre Lindenbaum118k
2
gravatar for Nicola Casiraghi
3.8 years ago by
Trento, IT
Nicola Casiraghi440 wrote:

If you inspect the run_disopred.pl file at line 20 you can find:

## IMPORTANT: Changing these flags will alter the behaviour of blastpgp
## You may want to use -a n to speed-up the search using n processors, if available
my $PSIBLAST_PAR = "-a 1 -b 0 -j 3 -h 0.001";

So, at least this part of the computation can be performed using more than one processor by properly setting parameter a.

ADD COMMENTlink modified 3.8 years ago • written 3.8 years ago by Nicola Casiraghi440

Wow I think that's what I need! I looked through all the files but I guess I didn't see this part.... thanks a lot! :)

ADD REPLYlink written 3.8 years ago by mparadis9240
Please log in to add an answer.

Help
Access

Use of this site constitutes acceptance of our User Agreement and Privacy Policy.
Powered by Biostar version 2.3.0
Traffic: 773 users visited in the last hour