The protein I have selected from my screening didn't have a known 3D structure, so I had to generate it. Some of the servers I used are - Phyre, I-TASSER, Quark, CABS-fold, RaptorX, Robetta.. but almost all of them didn't give the correct model. I can tell this because the common unstructured regions as predicted by IUpred, ANCHOR and Disopred were predicted to be helices in the models. Finally I decided to take only the fragment (intrinsically disordered protein) which contained my B-Cell antigen because I have to look at antigen-antibody simulation.. so the whole protein structure doesn't matter anyway.
But now I need to design/generate an antibody for my antigen since there have been no experiments on this protein and no known antibody is known.
You can use PEP-FOLD3. I do not understand how disorder prediction methods can predict helices
Hi,
I think you can use the module present in Biovia (Accelrys) Discovery studio software to design antibody (http://accelrys.com/products/collaborative-science/biovia-discovery-studio/antibody-modeling.html). You can also do a MHC Class I and II binding predictions with your antigen to check its affinity towards the HLA molecules if required.
regards,
Navya
@Phillippe
Sir,
The disorder prediction methods did not predict helices.
The protein 3D structure prediction methods predicted helices at the places which were shown to be disordered regions by IUpred, ANCHOR and Disopred.
Thus the models were rejected by me.
Any suggestion on how to design the antibody In-silico from the antigen sequence itself would be highly appreciated.
Regards,
Sandeep
@ Navya Raj
Hello,
Thank you. I'll look through the server right away.
Can you please help me with another thing?
IEDB website has a method called PIGS in B-cell epitope prediction tools.
http://tools.iedb.org/pigs/
This server requires a heavy chain and light chain sequences for the antibody structure prediction. So how do I bring in the above sequences?
Sincerely,
Sandeep
To ab-initio predict the sequence of antibody that binds a specific peptide is extremely difficult. Even with targets of known structure and binderes based on a more rigid scaffold, the best protein design labs only get only a fraction of the designed proteins to bind, with very moderate affinities (high micromolar, compared to low nanomolar for binders generated experimentally.
If you need an antibody for an experiment, and are not in the business of developing improved protein design and prediction methods, you are better off to let a mouse generate your antibody (monoclonal antibodies) or use in-vitro selection and affinity maturation methods (e.g. phage display).
- Badges
- Science topic
- Similar topics
- Biological Science
- Biochemistry
- Biomolecules
- Peptides