Re: [modeller_usage] Error when using dope_loopmodel
To:
Subject: Re: [modeller_usage] Error when using dope_loopmodel
From: Yun He <>
Date: Wed, 7 May 2008 19:17:17 +0800
Organization: Institute of Biophysics Chinese Academy of Sciences
Hi,
I encounter this error too ...
It works on my Debian Linux x86,
but it fails on my Fedora Linux x86_64 .
I don't know why ..
Yun
On 2008年3月18日 星期二, Modeller Caretaker wrote:
> Florian Odronitz wrote:
> ...
>
> > Please note that I used the dope_loopmodel class instead of loopmodel
> > as indicated in the manual.
> > When I run the script, it exits with the following message:
>
> ...
>
> > check_inf__E> Atom 4293 has out-of-range coordinates (usually infinity).
> > The objective function can thus not be calculated.
> >
> > Using dopehr_loopmodel gives the same error whereas the "old"
> > loopmodel works.
> > Here is the Atom record from the PDB file:
> >
> > ATOM 4293 N LEU 535 10.568 18.659 81.951 1.00103.06 1SG4294
> >
> > The coordinates do not seem to be out of range. I also put a blank in
> > the middle ("1.00 103.06") to avoid problems with the parser, but to
> > no avail.
>
> The error refers to the current set of coordinates (during optimization)
> - not the initial PDB coordinates. And don't put a blank into the PDB
> file - PDB files are fixed-column format, so it doesn't matter that
> there's no space between 1.00 and 103.06 (in fact, your "fixed" PDB file
> is now invalid).
>
> An atomic coordinate at infinity is usually a symptom of large forces in
> your system. These can occur if you have a bad initial model (e.g. two
> atoms superposed or very close). It is more likely to occur with DOPE
> loop modeling than regular loop modeling since the former includes
> GB/SA. (The loop modeling potentials themselves, while they are
> atomistic pair potentials, plateau as r tends to 0, unlike electrostatic
> potentials which tend to infinity.)
>
> > Am I doing something wrong here or is this a bug?
>
> Could be either. You can try a different initial model if your initial
> model is very bad. Otherwise, send me your input files and I'll try to
> reproduce your problem, to determine whether it's a bug in the GB/SA code.
>
> Ben Webb, Modeller Caretaker
--
Yun He Ph.D.
National Laboratory of Biomacromolecules
Institute of Biophysics, Chinese Academy of Sciences
15 Datun Road, Chaoyang District
Beijing 100101
China
Tel: +86 010 6488 8487
E-mail:
or