each of the loop copies doesn't see the other copies- what this means
in practice for PME calculations is that you need to do a correction
for every atom pair that should not interact. For very large LES
regions (many atoms per copy) this leads to large numbers of
corrections, as you have seen. you could change the code limits, but
it's likely to also be inefficient. multiple regions leads to better
averaging anyway- so it might be better to make separate copies of
each loop.
I should point out, though, that modeling a 50-residue loop isn't
likely going to work well with any MD based approach. This is the size
of a small protein. You probably want to use homology or other
fragment based methods.
On Sun, Sep 26, 2010 at 10:34 AM, Simon Becker
<simon.becker.uni-konstanz.de> wrote:
> Hi all,
>
> I'm working with a small protein bound to a transmembrane receptor.
> Unfortunately we are unable to crystallize the receptor alone, most
> likely because the long loops connecting the tm-helices are interfering
> with the crystal packing.
> To produce a relaxed structure of the receptor alone I considered using
> LES on these loops as the tm-region should be quite rigid. I put all
> interesting loops into one subspace and made 5 copies, like that:
>
> omas
> spac numc=5 pick #mon 404 445 | #mon 600 650 | #mon 542 576 | #mon 479
> 503 done
>
> is this even smart?
> Are there methods better suited for this?
>
> Addless however finished without errors after I increased a few
> parameters and recompiled.
>
> sander.LES.MPI stops with this error:
>
> LES parameters were found
> EXCEEDED MAXLESADJ!
>
> I found that maxlesadj is a prameter set in several .f and .h files to
> 3.000.000. This looks sufficient for my task, and I'm reluctant to
> temper with it.
> Does this mean there is a problem with my inputfiles?
> Is there any documentation concerning the les flags in the prmtop file?
>
> help please!
> simon
>
> _______________________________________________
> 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 Wed Sep 29 2010 - 13:00:03 PDT
