You are here

RosettaDesign parameters

5 posts / 0 new
Last post
RosettaDesign parameters
#1

Hello,

I am attempting to use RosettaDesign for the first time. I would like to identify mutations that will yield enhanced packing in the core of a 234 residue protein. In the manner of Dantas et al, JMB (2003) 332, 449-460, I would like to do 2 rounds of optimization: during the first round, I would like to allow the selected amino acids in the protein core to change into any other amino acid (except for CYS and GLY). During the second round, I would like to restrict the amino acid selection to those that were found in the first round.

For round 1, I have setup a resfile that contains amino acids in the core with the option ALLAA. I am using the following command:

rosetta -design -s acdt.pdb -fixbb -profile -ex1 -ex1aro -ex2 -ex2aro_only -ndruns 1000 -resfile reslist.res -pdbout acdt -scorefile acdt -no_new_CG -fa_input -try_both_his_tautomers -fix_disulf disulf.txt -norepack_disulf

However, I'm having difficulties with knowing if this set of keywords will yield the best results:

In the documentation, there was mention of using -use_aw and –use_bw. Is one of these keywords typically used in these types of design runs? The particular protein I’m studying has many beta sheets, so I’d assume that –use_bw would give better scores, but with the above keywords + -use_bw, I get approximately the same scores.

In addition, if I run the command with a resfile containing NATRO for all amino acids for 1 trial, the score is -234. However, for all 1,000 mutants generated during round 1, the score is always >= -227 and the acceptance is always 0.00. Is there something important that I’m missing, such as repack or mcmin_trials? Should I expect to achieve lower scores than the wild type protein during round 1?

Any help and advice would be greatly appreciated!

Thanks very much, Gregg

Tue, 2008-07-01 18:03
gbeckham

1. rosetta -design -s acdt.pdb -fixbb -profile -ex1 -ex1aro -ex2 -ex2aro_only -ndruns 1000 -resfile reslist.res -pdbout acdt -scorefile acdt -no_new_CG -fa_input -try_both_his_tautomers -fix_disulf disulf.txt -norepack_disulf
Do you have cys in your protein?The command line looks really
complicated. I would start with a simple one first and if it doesn't
work out well, then try to play with it.Try this first
rosetta -design -s acdt.pdb -fixbb -ex1 -ex2 -ex2aro_only
-ndruns 100 -resfile reslist.res -try_both_his_tautomers

2. The particular protein I'm studying has many beta sheets, so I'd assume that –use_bw would give better scores, but with the above keywords + -use_bw, I get approximately the same scores.

That suggest something is wrong. Usually when you use use_bw flag, the scores will be different. But actually, you don't have to use use_bw flag. Just don't use it at this moment, see what happens.
3. What you mean acceptance is always zero?Is the 1000 output structures
are exactly the same?

Huxz

> Hello,
>
> I am attempting to use RosettaDesign for the first time. I would like to identify mutations that will yield enhanced packing in the core of a 234 residue protein. In the manner of Dantas et al, JMB (2003) 332, 449-460, I would like to do 2 rounds of optimization: during the first round, I would like to allow the selected amino acids in the protein core to change into any other amino acid (except for CYS and GLY). During the second round, I would like to restrict the amino acid selection to those that were found in the first round.
>
> For round 1, I have setup a resfile that contains amino acids in the core with the option ALLAA. I am using the following command:
>
> rosetta -design -s acdt.pdb -fixbb -profile -ex1 -ex1aro -ex2 -ex2aro_only -ndruns 1000 -resfile reslist.res -pdbout acdt -scorefile acdt -no_new_CG -fa_input -try_both_his_tautomers -fix_disulf disulf.txt -norepack_disulf
>
> However, I'm having difficulties with knowing if this set of keywords will yield the best results:
>
> In the documentation, there was mention of using -use_aw and –use_bw. Is one of these keywords typically used in these types of design runs? The particular protein I’m studying has many beta sheets, so I’d assume that –use_bw would give better scores, but with the above keywords + -use_bw, I get approximately the same scores.
>
> In addition, if I run the command with a resfile containing NATRO for all amino acids for 1 trial, the score is -234. However, for all 1,000 mutants generated during round 1, the score is always >= -227 and the acceptance is always 0.00. Is there something important that I’m missing, such as repack or mcmin_trials? Should I expect to achieve lower scores than the wild type protein during round 1?
>
> Any help and advice would be greatly appreciated!
>
> Thanks very much, Gregg
>

Tue, 2008-07-08 09:42
yiliu

Thanks for your reply!

1. I have several disulfide bonds, so I was trying to preserve them with the -fix_disulf disulf.txt and -norepack_disulf options. As recommended, I tried:

"rosetta -design -s acdt.pdb -fixbb -ex1 -ex2 -ex2aro_only -ndruns 100 -resfile reslist.res -try_both_his_tautomers"

All of the scores in the pdb files are around -236 to -230, and the mutants are different, but I guess I am just confused on how to tell which of these mutants are significantly better, if any are.

2. By removing -use_bw and using soft_rep, I've found that I can get better scores, but they are still not much better than the wild type protein. On that note, how do you score the wild type protein typically? I took the crystal structure and ran design with NATRO in the res file, but I did that several times and get different scores between -236 and -233, which implies that Rosetta is manipulating the structure somehow. And should "score" be the metric by which I compare mutants to the wild type?

3. On the acceptance, if I look at, say acdt_0001.pdb, after the coordinates, I get:

ntrials: 1

%accepted: 0.00

rms_to_start: 2.49374551E-07

score: -235.91

In every generated pdb file, ntrials=1 and %accepted is always 0.00, but the mutants are not the same. When I try to place -ntrials 1000 (for instance) in the rosetta command line, the output is the same (i.e. ntrials=1, %accepted=0.00).

Again, thank you very much for your help with RosettaDesign! Cheers, Gregg

Wed, 2008-07-09 14:40
gbeckham

> 1. I have several disulfide bonds, so I was trying to preserve them with the -fix_disulf disulf.txt and -norepack_disulf options. As recommended, I tried:
> "rosetta -design -s acdt.pdb -fixbb -ex1 -ex2 -ex2aro_only -ndruns 100 -resfile reslist.res -try_both_his_tautomers"
> All of the scores in the pdb files are around -236 to -230, and the mutants are different, but I guess I am just confused on how to tell which of these mutants are significantly better, if any are.

You can simply read the "score" value in the output file. The lower is the better.
score: -352.65
env: -109.25 env_weight: 0
pair: -41.67 pair_weight: 0
vdw: 3.50 vdw_weight: 0

> 2. By removing -use_bw and using soft_rep, I've found that I can get better scores, but they are still not much better than the wild type protein. On that note, how do you score the wild type protein typically? I took the crystal structure and ran design with NATRO in the res file, but I did that several times and get different scores between -236 and -233, which implies that Rosetta is manipulating the structure somehow. And should "score" be the metric by which I compare mutants to the wild type?
If you use "NATRO" on every residue in the resfile, the scores should be the same. You can also use "-design -fixbb --natrot " in the command line without resfile to score native structure.

> 3. On the acceptance, if I look at, say acdt_0001.pdb, after the coordinates, I get:
>
> ntrials: 1
>
> %accepted: 0.00
>
> rms_to_start: 2.49374551E-07
>
> score: -235.91
>
> In every generated pdb file, ntrials=1 and %accepted is always 0.00, but the mutants are not the same. When I try to place -ntrials 1000 (for instance) in the rosetta command line, the output is the same (i.e. ntrials=1, %accepted=0.00).
%accepted is not very important here. It should be the monte_carlo counters/diagnostic. I do not know why it is always 0

Fri, 2008-09-05 11:11
yiliu

> 1. I have several disulfide bonds, so I was trying to preserve them with the -fix_disulf disulf.txt and -norepack_disulf options. As recommended, I tried:
> "rosetta -design -s acdt.pdb -fixbb -ex1 -ex2 -ex2aro_only -ndruns 100 -resfile reslist.res -try_both_his_tautomers"
> All of the scores in the pdb files are around -236 to -230, and the mutants are different, but I guess I am just confused on how to tell which of these mutants are significantly better, if any are.

The command line looks fine. You can simple read the "score" to evaluate the result, lower is better

> 2. By removing -use_bw and using soft_rep, I've found that I can get better scores, but they are still not much better than the wild type protein. On that note, how do you score the wild type protein typically? I took the crystal structure and ran design with NATRO in the res file, but I did that several times and get different scores between -236 and -233, which implies that Rosetta is manipulating the structure somehow. And should "score" be the metric by which I compare mutants to the wild type?

It should be the same as long as your resfile's format is correct. You can also use "-design -fixbb -natrot" command line to score native structure without using resfile

> 3. On the acceptance, if I look at, say acdt_0001.pdb, after the coordinates, I get:
ntrials: 1

%accepted: 0.00
>
> rms_to_start: 2.49374551E-07
>
> score: -235.91
>
> In every generated pdb file, ntrials=1 and %accepted is always 0.00, but the mutants are not the same. When I try to place -ntrials 1000 (for instance) in the rosetta command line, the output is the same (i.e. ntrials=1, %accepted=0.00).

I do not know why the %accepted is always 0 at the moment. You can ignore it.

Fri, 2008-09-05 11:18
yiliu