You are here

Loop Modeling with fixed backbone

10 posts / 0 new
Last post
Loop Modeling with fixed backbone
#1

Hi,

I wanted to model two tight turns in my protein keeping the backbone of the protein fixed. This was done to evaluate the change in the energy due to changes in loop sequence and also to look for low energy conformations. I wanted only some selected residues at those loop positions for which I made a resfile.

Here are the parameter for loop modeling :-

-loops:refine refine_kic
-loops:input_pdb 1gfl.pdb
-loops:loop_file loop_file.loop
-in:file:fullatom
-in:file:native 1gfl.pdb
-resfile : resfile
-ex1
-ex2
-loop:fix_natsc
-loop:refine_repack_cycles
-loop:max_outer_cycles 1
-loop:outer_cycles 1
-resfile: resfile
-out:file:fullatom
-out:overwrite
-out:prefix LP_MOD
-out:path ./
-nstruct 100
-mute core.util.prof ## don't show timing info
-mute core.io.database ## don't show database info

Loop File:-

LOOP 22 25 23 0 0
LOOP 115 118 116 0 0

Resfile :-

USE_INPUT_SC
start
23 A PIKAA HDSN
116 A PIKAA HDSN

Finally the results that I got were unexpected. Because, for the first loop Rosetta changed the last residue instead of the middle ones For eg. VNGH was changed to VNGR. I wanted to see the effect of replacing "N" with the residues mentioned in resfile. For the second loop , the sequence EDGT was changed to ENNA. Here last two were changed , I wanted the middle two only. Also, total_energy for all the structures was positive. It is the same protein (GFP, pdb id - 1GFL) that I am using for enzyme design.

loop_rms: 0.61525
total_energy: 625.813
chainbreak: 0.0481675

How could I specifically look for the changes in only the middle two residue of a four residue loop, without disturbing the stem of loop ??

Post Situation: 
Mon, 2011-12-19 05:20
bharat_46010

The primary problem is that you've put USE_INPUT_SC in your resfile for the "default" section. I am sure you meant either NATRO or NATAA, which mean "do not mutate this position" (with or without rotamer changes, respectively). USE_INPUT_SC only means that the input rotamer is retained as a possible output instead of forcibly changed to a Dunbrack-derived rotamer. So, fix your resfile to be something like:

NATAA
start
23 A PIKAA HDSN
116 A PIKAA HDSN

You may want to take a look at http://www.rosettacommons.org/manuals/archive/rosetta3.3_user_guide/file...

Mon, 2011-12-19 06:36
smlewis

So, I finally prepared the structure for Rosetta by using the following link and option :

relax.linuxgccrelease , constrain_relax_to_start_coords , relax:fast

Now I am getting the total_score of the modeled structures in negative. I generated some 100 structures and found that first structure has the most negative value, other structures values with a difference of -10 at max. I selected the first with lowest energy for loop modeling. I generated some 1000 structures and plotted the loop_rms & energy values. Plot is attached with this message. Now how to select the best from these structures. As 157 structures have same loop_rms and difference in total_energy is negligible , with least and max values as -553 and -552.

Wed, 2011-12-21 16:49
bharat_46010

If those two criteria are indistinguishable for your set, then you need to pick another thing to rank them on to find the "best". Ideally you'd use some sort of experimentally relevant info there (either data on what the loop does, or the idea of what you need it to do). You can try sorting by the other Rosetta scores, but since the total scores are similar that's unlikely to help. Rosetta is saying they're all the same, so you can probably just pick any of them and go with it.

Wed, 2011-12-21 17:01
smlewis

Thanks for your reply. I wanted to replace the native stem residues of the loop with the residue having same physio-chemical property. For eg. , to look for a better replacement for 'V' in VNGH I prepared the following resfile :-
NATAA
start
22 A PIKAA LIWF
23 A PIKAA HSDNG
24 A PIKAA HSDNG
25 A PIKAA HSPTY
116 A PIKAA HSDNG
117 A PIKAA HSDNG
118 A PIKAA HSPTY

I generated some 1000 models and all of them had 'V', not even a single structure had any residue changed as per the resfile. What does this mean and how could the replacement be done .

Thu, 2011-12-22 17:32
bharat_46010

What position are your trying to mutate away from V? (22, 23, 24, 25, 116, 117, or 118?)? What's your command line (to show the resfile is being used)? Is it mutating _any_ positions (if not, the resfile isn't being read)? You can deliberately introduce a bad line into the resfile to see if it crashes to check if it's reading the resfile.

Thu, 2011-12-22 18:37
smlewis

Here's the command line that I am using :-

./loopmodel.linuxgccrelease -database /home/pauling/Downloads/rosetta3.3_bundles/rosetta_database/ -loops:remodel no -loops:refine refine_kic -loops:fix_natsc -loops:input_pdb 1gfl.pdb -resfile resfile -in:file:fullatom -in:file:native 1gfl.pdb -loops:loop_file loop_file.loop -ex1 -ex2 -nstruct 1000 -out:file:fullatom -overwrite -out:prefix str

After adding a bad line (XXXXX) in resfile , I am getting the following error, which means that the file is being read :-
core.pack.task.ResfileReader: After token XXXXXXXXXXXXXXXX another token is expected.
ERROR:: Exit from: src/core/pack/task/ResfileReader.cc line: 1195

I also tried with loop remodeling for the same two loops(22-25, VNGH and 115-118, EGDT) with two different resfile :
Resfile 1 :
23 A PIKAA HSDN
117 A PIKAA HSDN

Resfile2 :

NATAA
start
22 A PIKAA LIWF
23 A PIKAA HSDNG
24 A PIKAA HSDNG
25 A PIKAA HSPTY
116 A PIKAA HSDNG
117 A PIKAA HSDNG
118 A PIKAA HSPTY

In the first case (Resfile1) the maximum occurrence is for residue N at position 117 and 25.

In the second case (Resfile2) the maximum occurrence is for residue G at 116, D at 117 and T at 118. For another loop the values are : Ile at position 22, N at position 23, G at position 24 , Y at position 25.

The plot for both the experiments have been attached. I want to know whether we need to get a funnel plot in this case also. How do I select the best loop sequence in the above two cases ??

Sat, 2011-12-24 23:55
bharat_46010

In general, funnel plots are RMSD versus score plots. Calculating RMSD implies that you know the real structure. Here, you are making mutations, which begs the question of what are you calculating RMSD against?

If you are trying to select loop sequences that will return a loop of the same backbone structure, but more stable, then you should select the sequence that is most common in the low-score, low-RMSD portion of the plot (or just the single lowest point).

If you are allowing a new backbone at the mutated positions, then the RMSD against the original loop is irrelevant, and you should just pick the best scoring model, or the model you think is most plausible for whatever other reason.

Thu, 2011-12-29 12:08
smlewis

Thanks for your reply. I want to ask whether Rosetta can help in predicting the change in protein folding rate after mutating such residues or not ??

Wed, 2012-01-04 21:16
bharat_46010

Hi,

I want know whether is it possible to add an extra residue during loop modeling in protein , which is not present in original sequence. For eg ATYG in original sequence and for model sequence ATXYG, where X is an extra residue.

Thanks
-----------
BHARAT

Fri, 2012-01-13 17:26
bharat_46010