I have constructed protein structure using I-Tasser serve. What evaluation criteria should be used for selection or rejection of the I-Tasser generated model?
I-Tasser results for my protein are as following:
c-score= -0.82, TM-score=0.61, RMSD=11
should I select protein model on these scores or should I evaluate model using other servers like Prosa, ERRAT, RamPage/Procheck.
Thanks
Because it is a model structure it is better to consult the ERRAT and other validation software. However You may wish to also consider the homology or the conserved fold (when homology is not good) in these group of proteins. Because It was seen that even if the homology is good the model structure sometime do not work for structure determination (a procedure called Molecular replacement by which a x-ray crystal structure could be solved). And after the same structure was determined using other method it was also found that there is not much structural deference only some loops or some residues are oriented in a differing conformation. So, most of the cases if the homology is high or if the protein belongs to a conserved fold group you can believe the structure to perform any biophysical or biochemical approach. Another approach could be to use Molecular dynamics (MD) simulations; During the MD simulation if you found that the r.m.s.d is varying too much of some regions of the Calpha backbone of the protein you can consider those are bad contacts,
HTH
Best criteria is RMSD and you should also check your build model at Verify_3D and GROMOS Energy Profile for each amino acids.
You can also simulate your model for 10 ns to analyse its stability at NPT
http://nihserver.mbi.ucla.edu/Verify_3D/
I recommend you to use ModEval server (http://modbase.compbio.ucsf.edu/evaluation/). Key scores here are predicted RMSD, z-score and predicted native overlap 3,5A.
Please find the article with description of this server in attachment.
Also it's a good idea to combine ModEval results with assessment by other tools performing deep sterochemical evaluation, such as MolProbity (http://molprobity.biochem.duke.edu/). Ramachandran plot should be checked with special attention.
My personal suggestion is ProSa2003 (Z-score), Ramachandran plot and ProCheck.
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology