Re: [AMBER] Is there a trick to fix a poor structural feature?

From: Jonathan Gough <jonathan.d.gough.gmail.com>
Date: Sat, 27 Apr 2013 20:39:08 -0400

Thank you everyone for their help.

Problem isn't solved yet. But here is a status update.

I took (my interpretation of) David Case's advice ("Finding out what is
going on is the first step towards a fix.")

The overall structure is of 4 proteins bound together(~20,000 atoms).
Here is what I have done:

1. I went back to the original structure from PDB, I Cut out a portion of
one (of the 4 proteins) and attempted to simulate it with and with out 2 of
the phosphoralated amino acids. there were No issues minimizing and heating
up and simulating either the phosphorylated or unphosphorylated versions.
conclusion - Phopho-parameters are ok.

2. From the original PDB, I treated each of the proteins separately I
replaced the phosphorylated amino acids with the unphosphorylated versions
(deleted the appropriate PO4 and let leap take care of the rest). There
there were a few residues missing (the loops I previously added) I simply
put in a TER line. Individually, the proteins gave no problems minimizing,
heating up, or doing an MD simulation.
Conclusion - individual proteins are OK

3. From the original PDB, I treated each of the proteins separately,
replaced the missing loops with a TER statement. Individually, the proteins
with the phosphates gave no problems minimizing, heating up, or doing an MD
simulation.
Conclusion - individual proteins with phos are OK

4. Took the 4 proteins together, no phos, TER statements instead of adding
in the Loops. very careful minimization (no overlaps within 0.95
angstroms), limited the Temp to 1K, took an infinitesimally small steps,
for a very limited times scale, random atoms (in my opinion) decided to
jump out towards infinity.
Conclusion - something is up when they are all together.
Next step is to look at 2 at a time....

I did try the YASARA server and It did give me a minimized structure for
the whole thing.
2 issues however.
  1 - It removed the phosphates then minimized. I'm really concerned about
adding them back in and having issues resulting from the conformational
changes and then having problems "fitting them in" without major issues. I
will try that out
  2 - I think I'm stubborn and want to understand what is going on
therefore I can better trouble shoot these things going forward.

again, thanks and if after reading my entire summary above have any insight
or ideas or suggestions, I would greatly appreciate it!

Best,
JOnathan






On Thu, Apr 25, 2013 at 11:04 AM, Symon Gathiaka <
smg0020.tigermail.auburn.edu> wrote:

> My 2 cents opinion:
>
> Try modelling the missing parts till you get a good structure first. This
> will save you a lot of trouble. I am not sure if swisspdb viewer does
> modeling of missing residues (i guess thats why you're getting bonds of 6
> angstrom between some residues).
>
> Once you insert the missing residues using the viewer, you have to use
> swiss-model to repair the structure. There are other tools for building the
> missing residues but swisspdb-viewer/swiss-model is the easiest to use.
>
> Symon
>
>
>
>
>
> WE ARE REACHING THE STAGE WHERE THE PROBLEMS WE MUST SOLVE ARE
> GOING TO BECOME INSOLUBLE WITHOUT COMPUTERS.
> I DO NOT FEAR COMPUTERS. I FEAR THE LACK OF THEM.
>
> -- ISSAC ASIMOV
>
> ________________________________________
> From: Jonathan Gough [jonathan.d.gough.gmail.com]
> Sent: Tuesday, April 23, 2013 1:47 PM
> To: AMBER Mailing List
> Subject: [AMBER] Is there a trick to fix a poor structural feature?
>
> I have a pdb (from an x-ray structure) that was missing several
> "loops"(they are floppy therefore were very disordered in the crystals).
> I used swisspdb viewer to add the residues back in but on a few of the
> "connections" it was a stretch.
>
> therefore leap warns me that
>
> WARNING: There is a bond of 6.802961 angstroms between:
> ------- .R<SER 363>.A<C 10> and .R<ARG 364>.A<N 1>
> WARNING: There is a bond of 3.723094 angstroms between:
> ------- .R<SER 990>.A<C 10> and .R<ARG 991>.A<N 1>
> WARNING: There is a bond of 4.437505 angstroms between:
> ------- .R<ALA 1197>.A<C 9> and .R<VAL 1198>.A<N 1>
>
> and (of course?) there aren't any improper torsion parameters and I get
> more warnings.
>
> the question is
>
> 1 - can I still use what I have and minimize (I think I have tried this
> several different ways and things go crazy)? If so any suggestions on how
> to go about doing it? (I tried/thought I could do it in implicit and
> explicit solvent followed by slow heating with heavy constraints and short
> times steps.)
>
> 2 - is there a "trick" to fixing what I have?
>
> 3 - is there a better (easier) way to build out missing residues denovo
> (letting amber build them by leaving out the residues does not work as it
> fills in the missing atoms in a straight line - not in a loop as I need)
>
> any insight or pointing in the right direction would be appreciated.
>
> Thanks,
> JOnathan
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
>
>
> _______________________________________________
> AMBER mailing list
> AMBER.ambermd.org
> http://lists.ambermd.org/mailman/listinfo/amber
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Sat Apr 27 2013 - 18:00:02 PDT
Custom Search