Thanks Ben,
Yes, I am acutely aware that our interface has got out-of-date.
I can add in these little fixes.
If there is sufficient interest, we could talk about updating to
the python version.
Cheers
Martyn
-----Original Message-----
From: CCP4 bulletin board on behalf of Ben Webb
Sent: Fri 11/2/2007 10:57 PM
To: [log in to unmask]
Subject: Re: [ccp4bb] Fwd: [ccp4bb] CCP4i and modeller integration
Jayashankar wrote:
> Dear friends and scientists,
>
> I want to ask about for an integration of modeller in the ccp4i
> interface. The interface in ccp4i is designed for modeller Version 6.2,
> we have by now modeller 9.2 , which is a lot different.
> Nevertheless I tried to enter the necessary data in ccp4i. I could not
> try the integration myself because I got no suitable test data set.
> Please can anybody inform me whether it is possible and the way to do that.
I am not especially familiar with ccp4i, but I am the lead developer for
Modeller, so can hopefully say something useful about that. ;)
True, the latest version of Modeller is 9v2, which is very different
from 6v2. However, if you are using 'old style' TOP scripts (which ccp4i
generates) then 9v2 will quite happily work with the same scripts that
work in 6v2.
I just tried this out with ccp4i (6.0.2) using the 'simple' modeling
example from the Modeller distribution (alignment.ali from
examples/automodel/). For the configure window for CCP4i, I just filled
in MODINSTALL to the same value as when I installed Modeller, left the
install key blank (not needed for 9v2) and filled in the program name as
'bin/mod9v2'. (The bin prefix is needed because CCP4i looks in the
MODINSTALL directory for the 9v2 script, not in the bin subdirectory
which is where it lives in Modeller 7 and later.)
To actually run Modeller, I put alignment.ali and pdb5fd1.ent in my
CCP4i project directory, pulled up the 'run Modeller' CCP4i window,
selected 'alignment file' for 'input sequence for structure to build
as', selected 'define TOP file' (seems not to work properly without
that, although I can't tell from the docs what it's supposed to do), put
in 'alignment.ali' for 'Align in', left 'Top in' blank, then added two
PDB files - one with code name '5fd1' and 'known in' 'pdb5fd1.ent', and
another with code name '1fdx' and a blank value for 'known in'. The Run
button should then run Modeller normally.
This should all be fine for 'no' refinement. For the 'fast' and 'full'
settings, you must edit ccp4-6.0.2/ccp4i/templates/modeller.com, and add
the line
1 SET PDB_EXT = ' '
immediately after the
1 INCLUDE
line. (This is because by default Modeller 7 and later write out files
with a .pdb extension, while ccp4i expects them to have no extension.)
The CCP4 guys may want to put this into the main CCP4 distribution,
since it will also work for Modeller 6 (where it is a no-op).
P.S. In at least one place in the CCP4 docs, the user is pointed to the
old Modeller page at Rockefeller, which no longer exists. That link
should go to http://salilab.org/modeller/ instead.
P.P.S. If there is interest from CCP4 users or developers in improving
the CCP4i/Modeller integration, I can certainly help out. For example,
TOP scripts are deprecated in Modeller 9 in favor of Python scripts, and
it would not be especially hard to change the script template to Python
(although of course this would then break backwards compatibility with
Modeller 4 and 6).
Ben
--
[log in to unmask] http://salilab.org/~ben/
"It is a capital mistake to theorize before one has data."
- Sir Arthur Conan Doyle
|
