I have predicted a protein structure based on peptide sequence, using SwissModel, Phyre2 and M4T Server ver. 3.0. All the three structures predicted by these software have minute variation. I have validated these structures using SAVes, of which SwissModel structure was found to be a more stable and appropriate one. Is there any other means for validating these structures further? Is the validation by SAVes sufficient enough to proceed further or do I need to validate the structure further?
I use Q-mean and Prosa to evaluate the predicted structure based on Z-scores.Try these two websites too.
As far as I know the results of swiss model are more reliable.
Good Luck
I have tried with many online tools and finaIy came to the conclusion that SwissModel icludes the best and most accurate quliity check tools ...like DFire, Anolea, Qmean6 and etc... my suggestion is to work with the Protein Structure & Model Assessment Tools in SwissModel.... its enough. It is important to know how to interpret the results and undestand the score.
there are several programs/servers (Vadar, Swiss Model, ModEval,...). if any domain is solved must be coincident with the structure. You mus see at least RMSD, energy, rmachandran plot (expected vs observed) and packing deffects
Try CING software, a bit more complicated than PROCHECK. Takes into acount packing quality and sidechain rotamers.
You can also use PDBSUM to validate your model which includes PROCHECK.
I'd say at this point you need to get your hands dirty and look at biochemical data.
What was the purpose of generating the model? Does it answer that aim? Is there any mutational data in the literature that could add weight to your structures? Can you generate your own? Either from binding assays or catalytic assays if it's an enzyme.
Does your structure help to answer any open questions regarding the protein of interest? Or can you generate functional and testable hypotheses from it?
Or finally, if you want to stick to computational methods, try some Molecular Dynamics simulations. If the structure just falls apart very quickly then it's unlikely to be stable.
Sorry for all the questions, but a predicted protein structure is always just a model and you need to test models in order to see if they are valid. Unfortunately, it's quite easy to make a believable model computationally. The best tests are the ones which could generate new biological insight.
You can take a look here for some suggestions on how to experimentally validate a structure short of fully solving it:
https://www.researchgate.net/post/Experimental_variables_that_can_be_calculated_from_model_protein_structures2
Jasmin Sutkovic, I also have problem of interpreting Anolea and Qmean6 results. Can we have an example please
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools