Homology modelling

Andrej Sali sali at tamika.harvard.edu
Wed Oct 12 17:44:12 EST 1994

In article <36nmhcINN2ng at bigbird.cc.williams.edu>  writes:
> Hello,
> 	Does anyone have any experience producing a very crude
> structural hypothesis from sequence data?  We have Sybyl, but I've
> been unable to get the Homology modeler to produce adequate results.
> The situation is this:  I have a sequence-structure coordinate file of
> a protein with > 85 % homology to the protein I want to make a
> structure for, but Sybyl complains about not having enough short
> regions of homology!  Are there any free programs that can do this?
> I've tried Sybyls built in structure-seq database, and constructed my
> own database with many similar proteins and just the resolved protein
> with homology.  Any suggestions or references would be appreciated,
> 				Max Nanao
> 				95mhn at cc.williams.edu
> 				nanao at ochre.mgh.harvard.edu

Depends on what exactly is meant by > 85% homology (% sequence identity?),  
but, by most definitions of homology in most situations like this, you  
should not have a problem to calculate quite an accurate homology model.  
At that homology level, there are likely to be only a few gaps in the  
alignment, the alignment is likely to be accurate, and the template's 3D  
structure is likely to be close to the actual target structure. I enclose  
below information about our homology modelling program MODELLER which is  
freely available to academics; you could use it to address your problem.  
If you have a template structure with 85% sequence identity to your  
target, the mainchain RMS error in the target model is likely to be around  
0.7A for at least 95% of the residues. Also, about 70% of the sidechain  
rotamers are likely to be predicted correctly (more for buried positions). 

Regards, Andrej

                        MODELLER 11, October 1, 1994
                    Copyright(c) 1989-1994 Andrej Sali           
                            All Rights Reserved                  

                           Written by Andrej Sali                
             Birkbeck College, University of London, London, UK  
                 Imperial Cancer Research Fund, London, UK       
                     Harvard University, Cambridge, USA          

Andrej Sali, Department of Chemistry, Harvard University, 12 Oxford St,
Cambridge, MA 02138, USA. Tel: (617) 495 1775. Fax: (617) 496 3204.
E-mail: sali at tammy.harvard.edu.


MODELLER is most frequently used for homology or comparative modeling
of protein three-dimensional structure: the user provides an alignment
of a sequence to be modeled with known related structures and MODELLER
will automatically calculate a full-atom model. More generally, MODELLER 
models protein 3D structure by satisfaction of spatial restraints 
(A. Sali & T.L. Blundell. J.Mol.Biol. 234, 779-815, 1993). In principle, 
the restraints can be derived from a number of different sources. These 
include homologous structures (comparative modeling), NMR experiments 
(NMR refinement), rules of secondary structure packing (combinatorial 
modeling), cross-linking experiments, fluorescence spectroscopy, image
reconstruction in electron microscopy, site-directed mutagenesis, 
intuition, residue-residue and atom-atom potentials of mean force, etc. 
The output of MODELLER is a 3D structure of a protein that satisfies 
these restraints as well as possible. The optimization is carried out 
by the variable target function procedure employing methods of conjugate 
gradients and molecular dynamics with simulated annealing. The program 
includes a 100-page manual. MODELLER is written in Fortran and is meant 
to run on a UNIX system.


MODELLER is available free of charge to non-profit institutions.

First, please use the anonymous ftp account on tammy.harvard.edu (IP to copy the files from the pub/modeller directory to
your computer.  Print and sign the license form (academic-license.ps)
and mail or fax it to Andrej Sali. You will then receive the
encryption key (MODELLER_KEY) with which you will be able to unpack
the encrypted distribution file (modeller11-exec.tar.Z.cr). See file
INSTALLATION for installation instructions.

MODELER is also available as part of QUANTA, a large program with many
other functionalities including interactive graphics, CHARMm, and 
X-PLOR. If you are interested in this commercial version please contact 
Ms. Jo Ellen Collins at Molecular Simulations Inc, 16 New England 
Executive Park, Burlington, MA 01803-5297, tel: (617) 229 9800, 
fax: (617) 229 9899, email: jcollins at msi.com.


src\                sources or executables for MODELLER and associated 
modlib\             libraries and data files for the programs;
scripts\            script files used to compile and use MODELLER;
doc\                MODELLER documentation;
Makefile            Makefile for compiling/installing MODELLER modules; 
                    used by the Install script;
modeller11.README   this file;
INSTALLATION        how to install MODELLER;
Install             compilation and installation script;
tests\              tests and examples;

More information about the Molmodel mailing list