I have performed Homology modeling using discovery studio with template identity 83%. When I have analysed my results with PROCHECK and WHATCHECK it gives 91 % residues in allowed regions in ramachandran plot and -4.88 Z score in proSA server. But during WHATCHECK and PROCHECK it is showing some error like in PROCHECK , (1) Residue properties: Max.deviation: 5.6 Bad contacts: 2 * Bond len/angle: 13.2 Morris et al class: 1 1 2 + 3 cis-peptides G-factors Dihedrals: -0.02 Covalent: -0.28 Overall: -0.11 (2) M/c bond angles: 92.7% within limits 7.3% highlighted 3 off graph
and in WHATCHECK (1)Missing unit cell information (2) missing symmetry information (3) abnormally short inter-atomic distances (4)Backbone conformation Z score very low.
Should I be aware about these problems? If so, what could be the significant steps to resolve these issues?
Although overall Goodness of your model is (G factor is -ve) good and you can use this model for your studies still further energy minimization may be opted to increase your model quality.
Very concise and good response dear Rajnikant. I would like to add in it that Ramachandran plot can tell a whole lot of story about structural stereochemistry. By far your structure has shown good score by keeping the 91% residues in allowed region. No homology model is perfect, there are always accounts of local entropy accumulations which affects badly in overall added score.
If you have not energy minimized your model yet, then I will suggest you to use around 60-100 steps of Steepest Descent. After this, do further energy minimization using conjugate gradient until the rate of change of energy between two steps don't drop down to 0.005kcal/mol.
If you feel like disturbances in side chain atoms stereochemistry then KobaMin server may be tried to better the structure.
http://csb.stanford.edu/kobamin/
I hope, I have explained comprehensively. I would like to listen more from you, I might be a little more help to you.
Yours Sincerely,
Dear Eshan, it might be interesting performing the same analysis on the structure you used as a template. You might find that some of the "problems" associated to your model have been "inherited " by the template structure they were built from. If that is the case I would worry less.
Thank you all for all your responses. Mazhar and Fabio, I did the same which you have advised and now my modeled protein is showing pretty well results. I have another query as i am trying to predict active site in my modeled structure but since last month Q site finder and pocket finder is not working and i am new in bioinformatics so can you please suggest me something for active site prediction the results of which could be published.
I am trying to validate a protein template into saves server. But it shows that the size of the template is too large. What should I do now?? How much time take the discovery software to make homology modelling.
use Procheck server
discover studio visualizer for Ramachandran plot.
- Badges
- Science topic
- Similar topics
- Filing