You are here

modelling loops and inserted domains

7 posts / 0 new
Last post
modelling loops and inserted domains
#1

Dear all

I would like to model inserted domains of unknown structure and large loops. I have been looking at the several loop modeling options and have a few queries

1) Most basically, is use of the loop protocols the way to go? The advantage is the use of the environment from the 'host' domain, but are any of the protocols capable of ab initio modeling of large domain insertions (50-100 residues)? The inserted domain could be modeled alone, but that modeling loses the context information from the host domain.

2) After some work, I have made got the template structures read in, albeit with this warning

WARNING:: end of pdb file reached: angle, secstruct, & res info not found

I then see this error
increase max_loop_depth in loops_ns.cc
40 41

I have tried editing loops_ns.cc, but with no effect. Does this need recompilation? Is there an alternative workaround? If not, what is the minimum unit that could be recompiled and what would be the command for that?

3) While making the loop file, the script looked for but didn't find

rosetta/rosetta_fragments/nnmake_database/vall_cst_coord.dat.2006-05-05

Checking again, this file isn't in rosetta_fragments-2.3.0.tgz. Should it be?

I could get the loops by substituting with vall_cst_coord.dat.2001-02-02. Will performance be much degraded by using this older, smaller file instead?

Thanks very much in advance for your help

Daniel

Fri, 2009-01-16 02:14
danzinho

1) 50-100 residues is pretty large but you can always try
2) Yes, you definitely need to re-compile the source code to get a new executable
3) yes, vall_cst_coord.dat.2001-02-02 will work. You need modify the scripts to find this file instead of old one.

> Dear all
>
> I would like to model inserted domains of unknown structure and large loops. I have been looking at the several loop modeling options and have a few queries
>
> 1) Most basically, is use of the loop protocols the way to go? The advantage is the use of the environment from the 'host' domain, but are any of the protocols capable of ab initio modeling of large domain insertions (50-100 residues)? The inserted domain could be modeled alone, but that modeling loses the context information from the host domain.
>
> 2) After some work, I have made got the template structures read in, albeit with this warning
>
> WARNING:: end of pdb file reached: angle, secstruct, & res info not found
>
> I then see this error
> increase max_loop_depth in loops_ns.cc
> 40 41
>
> I have tried editing loops_ns.cc, but with no effect. Does this need recompilation? Is there an alternative workaround? If not, what is the minimum unit that could be recompiled and what would be the command for that?
>
> 3) While making the loop file, the script looked for but didn't find
>
> rosetta/rosetta_fragments/nnmake_database/vall_cst_coord.dat.2006-05-05
>
> Checking again, this file isn't in rosetta_fragments-2.3.0.tgz. Should it be?
>
> I could get the loops by substituting with vall_cst_coord.dat.2001-02-02. Will performance be much degraded by using this older, smaller file instead?
>
>
> Thanks very much in advance for your help
>
> Daniel
>

Fri, 2009-01-16 11:15
yiliu

Hi

Thanks for the reply. Just to follow-up a bit

1) Is there any reason to prefer one loop modeling protocol over another for larger insertions? I could of course try them all, but perhaps logic or experience can help point the way.

2) Would I have to compile the whole she-bang? This is non-trivial and our systems person is away for a few weeks. What would be a suitable value for max_loop_depth? Is this related to the length of the loop in any way?

3) Will use of vall_cst_coord.dat.2001-02-02 lead to worse performance than if a more recent file were available? Could vall_cst_coord.dat.2006-05-05, or another recent file, be made available somehow?

Again, your help much appreciated.

Daniel

Mon, 2009-01-19 07:46
danzinho

It might be better that you can use domain insertion mode of rosetta.

> Dear all
>
> I would like to model inserted domains of unknown structure and large loops. I have been looking at the several loop modeling options and have a few queries
>
> 1) Most basically, is use of the loop protocols the way to go? The advantage is the use of the environment from the 'host' domain, but are any of the protocols capable of ab initio modeling of large domain insertions (50-100 residues)? The inserted domain could be modeled alone, but that modeling loses the context information from the host domain.
>
> 2) After some work, I have made got the template structures read in, albeit with this warning
>
> WARNING:: end of pdb file reached: angle, secstruct, & res info not found
>
> I then see this error
> increase max_loop_depth in loops_ns.cc
> 40 41
>
> I have tried editing loops_ns.cc, but with no effect. Does this need recompilation? Is there an alternative workaround? If not, what is the minimum unit that could be recompiled and what would be the command for that?
>
> 3) While making the loop file, the script looked for but didn't find
>
> rosetta/rosetta_fragments/nnmake_database/vall_cst_coord.dat.2006-05-05
>
> Checking again, this file isn't in rosetta_fragments-2.3.0.tgz. Should it be?
>
> I could get the loops by substituting with vall_cst_coord.dat.2001-02-02. Will performance be much degraded by using this older, smaller file instead?
>
>
> Thanks very much in advance for your help
>
> Daniel
>

Thu, 2009-03-05 09:40
yiliu

Hi, but how to use domain insertion mode to do loop modeling? In my situation I got 96 residues in the loop without structure.

> It might be better that you can use domain insertion mode of rosetta.
>
>
> > Dear all
> >
> > I would like to model inserted domains of unknown structure and large loops. I have been looking at the several loop modeling options and have a few queries
> >
> > 1) Most basically, is use of the loop protocols the way to go? The advantage is the use of the environment from the 'host' domain, but are any of the protocols capable of ab initio modeling of large domain insertions (50-100 residues)? The inserted domain could be modeled alone, but that modeling loses the context information from the host domain.
> >
> > 2) After some work, I have made got the template structures read in, albeit with this warning
> >
> > WARNING:: end of pdb file reached: angle, secstruct, & res info not found
> >
> > I then see this error
> > increase max_loop_depth in loops_ns.cc
> > 40 41
> >
> > I have tried editing loops_ns.cc, but with no effect. Does this need recompilation? Is there an alternative workaround? If not, what is the minimum unit that could be recompiled and what would be the command for that?
> >
> > 3) While making the loop file, the script looked for but didn't find
> >
> > rosetta/rosetta_fragments/nnmake_database/vall_cst_coord.dat.2006-05-05
> >
> > Checking again, this file isn't in rosetta_fragments-2.3.0.tgz. Should it be?
> >
> > I could get the loops by substituting with vall_cst_coord.dat.2001-02-02. Will performance be much degraded by using this older, smaller file instead?
> >
> >
> > Thanks very much in advance for your help
> >
> > Daniel
> >

Tue, 2010-03-02 05:29
wenchangyu2008

A suitable value for max_loop_depth is equal or greater than the max loop library size you are using (line count of the loop library file). Note: max_loop_depth is number of loops in your library, nothing to do with loop length. the latter is max_loop_size

Thu, 2009-05-07 11:41
vanita

But how to use domain insertion mode to do loop modeling? In my situation I got 96 residues in the loop without structure. Also I checked the options for loopmodeling, which also include abinitio: inside. I tried to model this long loop and actually I got several structures and I think this can be compared with experimental data.

But I still do not know whether it make sense or not to use loopmodeling do model this long long loop. Thanks for any comments and suggestions.

Best,
Wenchang

Fri, 2010-03-19 17:05
wenchangyu2008