Geno3D – release 2

Introduction :

The aim of Geno3D server is to make accessible to all biochemists and biologists an automated protein modelling Web server to generate protein 3D model. This server is available in the hope that it will be useful, the entire risk as to the quality and performance of the server is with you.

The result is only a model and must be considered carefully, it isn’t an experimental 3D structure!

References : Combet, C., Jambon, M., Deléage, G. & Geourjon, C., Geno3D an automated protein modelling Web server, Bioinformatics, 2002, 18, 213-214

Contact : Dr. Christophe Geourjon,, E-mail :

Improvement from release 1 :

-        PSI-Blast on non-redundant proteins sequences database are available.

-        Possibility to select template whit low identity level (until 20%, use it carefully).

-         Use prediction of secondary structure agreement to validate template selection and alignment (Geourjon, C., Combet, C., Blanchet, C. & Deléage G., 2001, Identification of related proteins with weak sequence identity using secondary structure information. Protein science, 10: 788-797).

-        Selection of multi-templates

-        Availability of model analysis tools in Web interface

-        All files are included in archive file which can be download for local use.

-        Include in superposition file with template ligand (geometrical position deduce from template)

 Availability and limitation in Geno3D :

 GENO3D is available :

-        At URL :

-        Geno3D Web server is provided and developed by "PBIL Lyon-Gerland" team. There are no restrictions on its use by non-profit institutions. Usage by and for commercial entities requires a license agreement (E-mail:

 Limitation :

-        Geno3D server release 2 generates model until 500 amino acids.


By default the modelling process is made using :

-        Database : NPSA 3D SEQUENCES AT 100% HOMOLOGY (from PDB)  

-        Until 3 run for similarity search with PSI-BLAST program

-        Generation of up to 3 models

 Strategy :

The strategy used in Geno3D is comparative protein structure modelling by spatial restraints (distances and dihedral) satisfaction.

Geno3D is most frequently used for homology or comparative protein structure modelling :

ü     the user provides sequence to be modelled (query) which is compared using PSI-BLAST method (Müller, A., MacCallum, R.M., and Sternberg, M.J.E. 1999, Benchmarking PSI-BLAST in genome annotation. J. Mol. Biol. 293: 1257-1271) against a protein sequence database issue from PDB (all entries by default).



ü     the user selects PDB entries as templates for molecular modelling. In this case, 1bep, 1ccl and 1bek are selected.


ü     for each template Geno3D server computes secondary structure prediction, display percent of agreement in secondary structure (geourjon et al., 2001), repartition of information from template on query sequence. In case of selection of multi-template the rmsd (root mean square deviation) between templates on carbon alpha is also displayed :

Structure analysis for chain number   1

- Sequence of this chain : 

- This chain was modelled using 3 templates :

  Template          Alignment (Clustalw format)     Secondary information (Sov)     Identity 
   pdb1bek_0          alignment_1_1.aln                    71.0%                     18.5%
   pdb1bep_0          alignment_1_2.aln                    71.0%                     18.5%
   pdb1ccl_0          alignment_1_3.aln                    69.2%                     18.1%
- Template at each amino acid position :

- Deviation between templates on this chain (Angstrom) :
                      1bek   1bep   1ccl
          1bek        0.00   0.11   0.22
          1bep        0.11   0.00   0.21
          1ccl        0.22   0.21   0.00
     . mean deviation :   0.181526
     . superimposed pdb file : super_temp_1.pdb
     . local deviation along chain sequence : jpeg  text
          (Reference : Amino acid model numbering)

Enter your e-mail address :

Choose number of model to generate :

Inter/intra restraints ratio :

Margin in distance restraints (Angstrum):

Margin in angle restraints (degree) :

Maximal number of distance restraints :

Save full template in superposition :


In this example, templates share only 18% identity with the query, but the secondary structure agreement is quite good (70%) so it’s reasonable to make a molecular modelling with this  template (see geourjon et al., 2001).  We choose to conserve the 3 templates, even if for 1ccl the Sov value is low (it’s why line is written in red). Using 3 templates gives more flexibility than with only 1 template, but at the opposite that may introduce conflicts, so use modelling multi-templates selection carefully :

v    You need to confirm this selection, and launch Geno3D computing after enter your E-mail. Geno3D job will be running as soon as possible, depend my on computer server availability (you will receive a mail when your job is starting).

v    If you want to change the template choice, you go back and select another.

ü     distances and dihedral angles restraints calculated from the alignment (generated by PSI-BLAST) with templates 3D structure are applied onto on the query sequence. For gaps, statistical restraints are used.

ü     this restraints are used as input for CNS software (Brunger, A.T., Adams, P.D., Clore, G.M., Delano, W.L., Gros P., Grosse-Kunstleve R.W., Jiang, J.S., Kuszewski, J., Nilges, M., Pannu, N.S., Read, R.J., Rice, L.M., Simonson, T. & Warren G.L. Crystallography & NMR : A new software suite for macromolecular structure determination. Acta Cryst., 1998, 54, 905-921)

ü     the output are 3D models that satisfy theses restraints as well as possible.

ü     at the end of the molecular modelling process you receive a mail which provide an internet address to molecular modelling results (available for seven days). You can use this result via internet or download an  archive file (archive.tar.Z) for analyse results locally on your computer also via a Web interface. This archive must be uncompressed with command :

v    on Unix : zcat archive.tar.Z|tar -xvf -

v    on PC (windows) : with WinZip software (shareware available for example on tucows server :

v    on Mac : with Stuffit expander software (Freeware available for example on tucows server :

ü     On the server, the main page is :

RESULTS OF MODELING 2002011712204724039


. model_1.pdb
. model_2.pdb
. model_3.pdb


ABOUT WHOLE MOLECULE (energy, ramachadran, ...) : report.txt  
ABOUT EACH CHAIN (deviation, restraints, comparison with each template...) :
- Chain 1 :


From this page you can download individual model generated, so you can display this 3D model using a software as rasmol (Freeware available at ). You can also retrieve to some information about modelling process :

v    file report.txt : Information about whole molecule, ie information about energy, Ramachandran diagram (computing using PROCHECK software by Roman A. Laskowski, deviation between models, number of restraints deduce from template, and number of restraints which aren’t respected in models (violations). When a parameter is not satisfactory the line is written in red color.



model  1 :   -8850.40
model  2 :   -8712.33
model  3 :    2859.57
Energetical value
You can select model 1 or model 2 
       Model1 Model2 Model3
Model1  0.00   1.20   1.29
Model2  1.20   0.00   1.24
Model3  1.29   1.24   0.00 
Deviation between models on CA
(rmsd in Angstrum)
Model1       91.2
Model2       91.3
Model3       94.3

Ramachandran repartition
(% percent of residus in favorable region)

Superposition of 3 models generated

Restraints repartition

Violations repartition

v    file report_1.txt (and for a dimmer report_2.txt) : Information about each chain, ie deviation between models on this chain, number of restraints deduced from template, and number of restraints which aren’t respected in models (violations). Deviation between this chain and template used to build this chain.


      1bep Mod1 Mod2 Mod3
1bep  0.00 1.38 1.23 1.46
Mod1  1.38 0.00 1.20 1.29
Mod2  1.23 1.20 0.00 1.24
Mod3  1.46 1.29 1.24 0.00
Deviation between one template and models
Superposition (left figure):
- Red : template pdb1bep
- Yellow : Heme (from pdb1bep)
- Blue : model 1 to 3

Illustrations of analysis tools usage :

1 Using some templates for modelling the same protein chain. For example when a protein region was unresolved in a 3D structure, but resolved in another structure.



In pdb1jsq structure 2 regions are unresolved (in yellow color 207-238 and  343-420). The second region correspond to the “ATP binding domain” has been resolved in a structure of a homologous protein (code pdb1jj7). So using this two templates, it is possible to built a model for an ABC tranporter family member.

Future developments :

-        Modelling of dimers  

-        Include ligand in molecular modelling process  

-        Upload personal alignment

User : public@ Last modification time : Wed Feb 13 08:55:06 2002. Current time : Tue Feb 21 22:10:40 2017