I have a protein sequence whose str is to be predicted. This protein is a homodimer, what is the best way to model its dimer:
1. Shall I use Modeller software for homology modeling to make its dimer directly based upon its template that as homodimer? If yes, shall I go for dynamic study of this dimer?
OR
2. shall I go for homology modeling of one subunit using Modeller and then for docking by providing this homology model to make the dimer?
If template structure is already available as homodimer then you can directly model homodimer using modeller.
If i amy another question to this , if template structure is not known then what should be the process?
If template is not known then u should go for fold recognition and threading...use Genthreader
We can go for Homology modelling by using Swiss workbench , MODBASE,
And Docking is done using HEX as HEX can provide Docking for flexible , rigid ligand prot , prot prot Dock.
as compared to AUTODOCK and iGEMDOCK
Actually, that depends on your structure of the homodimer. Some homodimer is not 2-fold symmetric ...
Use the symdock server:
http://bioinfo3d.cs.tau.ac.il/SymmDock/
and the ClusPro 2.0 servers
http://cluspro.bu.edu/login.php
(you need to register)
In ClusPro you need to declare the same monomer both as receptor and ligand
You will get a lot of models in both methods.
If you have some experience in bioinformatics you can also use Rosetta Dock. Put two copies of your monomer close to each other using any structure visualization/manipulation program (VMD, PyMol, Chimera, NOC, even Swissprot can do it). Put it to dock in Rosetta dock and generate many decoys.
Compare the best results on each method (in cluspro there are four manners of deciding which models are best). And select those coincident results in all strategies (they may still be many).
Next you need to use experimental information to tell apart the true one. Here you need to be creative. For instance, most of the time, you do not want the active site to be hidden behind the protein-protein interface. Go to the sequence databases (NBCI, EMBL, Expasy, etc) and see if there are annotated conserved sites. You may find that information useful for discrimination. Use all possible data you can gather and relate to structure.
It is tough work and kind of uncertain, but that is as good as it gets.
Best wishes,
Rogelio
You may use modeller, if you a can identify a dimeric homologue of your protein. Otherwise, modeler will only produce the model for the monomer.
Modeller does not have tools to form the dimer. In theory, modeller ca do MD simulations, however, its features are limited and only short-time, simulations with constraints are recommended, usually to improve loops or small parts of a model.
Thanks a lot every one for valuable suggestions. I did this in MODELLER as the template of the target is available.
Regards,
Nutan
- Badges
- Science topic
- Similar topics
- Filing