Workflow
log in

Advanced search

Message boards : Number crunching : Workflow

Author Message
Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1914 - Posted: 5 Dec 2009, 4:38:39 UTC

This thread explains the status of our workflow.
____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1916 - Posted: 5 Dec 2009, 15:33:47 UTC - in response to Message 1914.
Last modified: 5 Dec 2009, 16:01:02 UTC

working on the workflow scripts this morning. will spend a few hours with schedular down.
____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1917 - Posted: 5 Dec 2009, 17:26:17 UTC - in response to Message 1916.

Now I am allowing the first two scripts to run continuously so we can generate some results for the extension script
____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1929 - Posted: 8 Dec 2009, 4:22:00 UTC - in response to Message 1917.

I updated the workflow to include Position-Restrained Dynamics
This step will help relax the protein ligand complex before we send it out as workunit. This tutorial details some of the steps we are taking.

Minimization -> Position-Restrained Dynamics -> Molecular Dynamics
____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1933 - Posted: 9 Dec 2009, 2:22:46 UTC - in response to Message 1929.

Now we are experimenting with some additional changes to the workflow.

Constrained Minimization -> Minimization -> Position-Restrained Dynamics -> Molecular Dynamics

____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1936 - Posted: 9 Dec 2009, 15:09:40 UTC - in response to Message 1933.
Last modified: 9 Dec 2009, 15:10:11 UTC

I have a very small sample of results to analyze (5), even though I have no errors, I cannot make any conclusions at this time. Currently the time it takes to generate workunits can vary depending on how much I minimize the the structures. Currently it takes about an hour to minimize on our server. I want to see a sample of some 10 workunits before I draw conclusions.

Would it make more sense to do the minimizations steps as workunits? Maybe, but it could also have errors that we have the benefit of catching on the server.


I also modified a feature in our parameter files, I set the random seed parameter so each result would use the same random seed for a given workunit. It would generate a new random seed number for each workunit.
____________

Di
Avatar
Send message
Joined: 13 Nov 09
Posts: 13
Credit: 36,826
RAC: 0
Russia
Message 1937 - Posted: 9 Dec 2009, 16:41:25 UTC
Last modified: 9 Dec 2009, 17:03:05 UTC

Jack can you explain with more details what is this steps means(Constrained Minimization -> Minimization). Why do you need to minimized a structures and if it takes one hour on the server than you should include it in wu's I think...
____________
From: Moscow, Russia | DrugDiscovery,Milkyway/Collatz,FreeHal,Majestic12 at E7300 3.2Ghz, Radeon 3850 HIS turbo GHz | Team: Russia

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1938 - Posted: 9 Dec 2009, 17:08:13 UTC - in response to Message 1937.

mdrun is sensative to high energy interactions. It creates a great deal of random disorder. The software cannot make reasonable models of such a system and it is useless for modeling biological systems. If the model is under these conditions, it means you did something wrong in preparing. A way to overcome this issue is to perform a minimization step which effectively cools down the molecules so they would behave in a reasonable manner. Ideally you would have them minimized to the maximium force on any atom being bellow 10, this is the parameter

emtol: (10.0) [kJ mol-1 nm-1]
the minimization is converged when the maximum force is smaller than this value

We aim to relax these molecules by bringing down the energy and forces on the individual atoms AND to restrain the distance between bonds. It helps the molecules explore lower and lower energy levels.

For more details on the actual steps used by the software, you can read this tutorial
http://eugen.leitl.org/chem/kerrigje/pdf_files/trp_drug_tutor.pdf
____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1939 - Posted: 9 Dec 2009, 17:10:58 UTC - in response to Message 1938.

I added the extension script again and that caused some errors. I turned it off for now and will wait until some of the older workunits pass which did not have the new minimization steps.
____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1940 - Posted: 9 Dec 2009, 17:18:40 UTC - in response to Message 1939.

keep an eye on workunits that have this name
nsteps_5000_int_0_-_1000_bt_ [etc etc etc]
these are the ones that are using the new minimization steps. Out of 6 workunits and 18 results I have not seen any errors
____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1943 - Posted: 9 Dec 2009, 19:29:41 UTC - in response to Message 1940.
Last modified: 9 Dec 2009, 19:31:17 UTC

a new change to the workflow. Now I made the number of restrained protein dynamics steps 500 which is a 1 ps simulation.

so here is how it is

restrained minimize until fmax < 1000 or after 500 steps
unrestrained minimize until fmax < 10 or after 10,000 minimization steps
restrained protein molecular dynamics 1 ps or 500 steps

Generally every ps or 500 steps take a minute to simulate. Under our current parameters it takes 500 steps to simulate 1 ps.
It takes 15 minutes to prep each run under these conditions. That is better than an hour. If it works under these conditions. How about changing the unrestrained minimization to 5,000?
____________

Profile Tim Turner
Avatar
Send message
Joined: 1 May 09
Posts: 570
Credit: 184,322
RAC: 0
United States
Message 1944 - Posted: 10 Dec 2009, 5:57:30 UTC - in response to Message 1943.

a new change to the workflow. Now I made the number of restrained protein dynamics steps 500 which is a 1 ps simulation.

so here is how it is

restrained minimize until fmax < 1000 or after 500 steps
unrestrained minimize until fmax < 10 or after 10,000 minimization steps
restrained protein molecular dynamics 1 ps or 500 steps

Generally every ps or 500 steps take a minute to simulate. Under our current parameters it takes 500 steps to simulate 1 ps.
It takes 15 minutes to prep each run under these conditions. That is better than an hour. If it works under these conditions. How about changing the unrestrained minimization to 5,000?


7.31 mdrun a bust, all error out within 40 seconds.

____________
Tim Turner
Public Relations Admin
Secunia PSI: http://secunia.com/vulnerability_scanning/personal/
If you need help via voice or Convo; PM me and i will give you details on where i will be; Teamspeak, Yahoo Messenger, or Skype.

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1945 - Posted: 10 Dec 2009, 11:33:55 UTC - in response to Message 1944.

Its not the version but the way we are generating the work. The extension scripts are failing to generate good work.
____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1946 - Posted: 10 Dec 2009, 12:08:22 UTC - in response to Message 1945.

extension works on my computer, but fails for everyone else
____________

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1947 - Posted: 10 Dec 2009, 12:46:43 UTC - in response to Message 1946.

I came up with a modification to the extension script. These are the steps you need to take to extend a simulation using gromacs

tpbconv -s previous.tpr -extend timetoextendby -o next.tpr
mdrun -s next.tpr -cpi previous.cpt

If I want to extend a simulation by 1 ps I would do

tpbconv -s previous.tpr -extend 1 -o next.tpr
mdrun -s next.tpr -cpi previous.cpt

What I will do is run a 0.5 ps extension before sending to workunits

http://www.gromacs.org/Documentation/How-tos/Extending_Simulations
____________

Profile Tim Turner
Avatar
Send message
Joined: 1 May 09
Posts: 570
Credit: 184,322
RAC: 0
United States
Message 1951 - Posted: 10 Dec 2009, 19:15:00 UTC - in response to Message 1945.

Its not the version but the way we are generating the work. The extension scripts are failing to generate good work.



sorry, i had a drink last night, it makes me loopy! and jump to conclusions!
____________
Tim Turner
Public Relations Admin
Secunia PSI: http://secunia.com/vulnerability_scanning/personal/
If you need help via voice or Convo; PM me and i will give you details on where i will be; Teamspeak, Yahoo Messenger, or Skype.

Profile Jack Shultz
Avatar
Send message
Joined: 10 Apr 09
Posts: 503
Credit: 120,150
RAC: 0
United States
Message 1952 - Posted: 11 Dec 2009, 0:49:40 UTC - in response to Message 1951.

lots of errors. I think I should do the testing on our test group now
____________

Message boards : Number crunching : Workflow


Main page · Your account · Message boards


Copyright © 2014 DRUGDISCOVERY@HOME