I would like to run multiple instances of rosetta on a cluster in order to speed
up decoy generation.
How can I ensure that each of the decoys is original so that the conformational
space is searched without duplicating decoys?
Does rosetta have a way of creating a random seed and if so how does this alter
the decoy generation?
Thanks
Mathew Smith
Hi Mathew,
You'll want to give each instance of rosetta a different random seed.
To do this, you'll use two flags in your command line: -constant_seed and -jran
The integer that follows the -jran flag should be between 1 million and 4 million.
Best,
Andrew
> I would like to run multiple instances of rosetta on a cluster in order to speed
> up decoy generation.
> How can I ensure that each of the decoys is original so that the conformational
> space is searched without duplicating decoys?
> Does rosetta have a way of creating a random seed and if so how does this alter
> the decoy generation?
>
> Thanks
>
> Mathew Smith
Hi,
What about -seed_offset option? Can it be used instead of -constant_seed -jran ? Is there any difference in the results if I use -seed_offset ?
What -seed_offset actually does? When I run two rosetta jobs with "-seed_offset 1" the random seed reported in the standard output is nevertheless different ...
Thanks,
Jan Kosinski