You are here

rna_denovo and minize_rna

3 posts / 0 new
Last post
rna_denovo and minize_rna
#1

Dear all,
I'm impressed with Rosetta for the generation of small RNA structures (nt ~ 20) using rna_denovo in combination with the -params_file and -minimize_rna options. When performing a number of runs to select the best scoring one I've noticed there's a correlation between the score of the unminimized structures and the scores after optimization. To speed things up I would like to first generate a larger number of unminimized structures and from this set pick the best scoring ones for further optimization. I'v tried this using rna_minimize but this doesn't seem to work: the resulting scores are much higher than what I get in a run with the minimize_rna flag. Am I missing something here?
Kind regards,
Frits.

Post Situation: 
Wed, 2013-07-10 00:22
frits

Hi Frits,

Were you using any extra options in the denovo protocol such as -relax_rna or -simple relax? Do you have any chemical shift data you are using?

There looks to be extra minimization if chemical shift data is present, as well as an extra loop closing step if using the minimize_rna option vs. the rna_minimizer.

The scores should still be good the way your using it, but it looks like the only way to get the full behaviour is to use the minimize_rna option.

Wed, 2013-07-10 08:35
jadolfbr

Hi,
I did try the vary_geometry options but wasn't wild about it. As constraints I use 2D constraints from ViennaRNA, which helps a lot. What I'm aiming at is to stop rna_denovo after the global minimization, and run the local minimization on the structure that score best in the first step. Actually, I must have done something wrong the first time I tried it because now it seems to work find, i.e.
rna_denovo -fasta test.fasta -database $ROSETTADB -nstruct 10 -out::file::silent -out::file::silent test0.rosetta.out -params_file test.rwc -minimize_rna -score:weights rna_lores.wts
gives me comparable results to
rna_denovo -fasta test.fasta -database $ROSETTADB -nstruct 100 -out::file::silent -out::file::silent test0.rosetta.out -params_file test.rwc -score:weights rna_lores.wts
followed by
rna_minimize -database $ROSETTADB -s S_0000nn.pdb -out:file:silent score2.out
The nice thing is that it seems that I only have to minimize a selection of the S_0000nn-s - the ones that score best in the first step - to get the best scoring structure.
Thanks for the reply,
Frits.

Fri, 2013-07-12 06:07
frits