Run it in Phyre2 or get yourself an accelerator-powered x-ray diffractometer, 800 NMR, or possibly a quartz crystal microbalance with dissipation monitoring or a dual-polarization interferometer; also a cryo-em would do the trick.

@vincent paul schmitz i dont have the experimental structure. I have the predicted structure

Dear Shilpa Sarkar ,

You mentioned I-Tasser. Or did you used GPCR-I-TASSER? See:

https://zhanglab.ccmb.med.umich.edu/GPCR-I-TASSER/

See for details:

https://zhanglab.ccmb.med.umich.edu/papers/2015_11.pdf

If you have a structure prediction with a low RMSD and a TM-score higher than 0.5 (and closer to 1.00 the better it is) then you can assume you have a good and probably reliable structure.

As mentioned and meant by Vincent Paul Schmitz the ultimate control on whether your structure is really ‘realistic’ can only be determined experimentally.

But I-Tasser and specifically GPCR-I-Tasser is basically the best and most state-of-the-art prediction program that proved to give results that are often remarkably close to (experimentally) known structures.

If your question implies what kind of bioinformatics can I perform to check my structure than you might consider matters like secondary structure predictions in order to see if for example helical regions in your predicted structure correspond. There are multiple secondary structure programs like PsiPred, JPred, SOPMA and so on.

The same is true for the transmembrane regions. Specialised programs are out there that pretty accurate pinpoint the correct transmembrane regions in a membrane protein. Look for programs/servers like TMHMM, TOPCONS and the recently developed online prediction tool MemBrain:

http://www.csbio.sjtu.edu.cn/bioinf/MemBrain/

Hope this is what you are looking for.

Best regards.

Rob Keller thank you sir. actually i was trying to validate my structure using SAVES but verify 3D shows a very low score. It also showed be some of the bond angles are out of range. How to correct those?

Dear Shilpa Sarkar ,

Please keep in mind that validation does not tell you how much the model is close to reality, but the validation algorithms take into account matters like certain geometrical parameters, topological and whether energy requirements meet their criteria.

Personally I would leave it where I-Tasser tells you how confident you can be about the prediction and have a look at some of the suggestions I made.

Still if you feel that you want to look at matters like disturbances in side chain atoms stereochemistry then the KobaMin server may be tried to improve the structure:

http://chopra-modules.science.purdue.edu/modules/kobamin/html/

Good luck with your work.

@Rob keller thank you sir. I have tried refining the model. The best verify 3d score shows that 55% amino acid has a score >0.2. Is it a big problem. I know it should be artist 80%

Ultimately, you cannot fully validate an in silico predicted structural model. "Predicted structures" are generally not trusted and not taken seriously unless there is a clear structure-function relationship which can validate the prediction experimentally or if there is a close structural homologue.

Also, in terms of energy minimisation - the I-Tasser website states the following:

"The decoys generated in the second simulations are then clustered and the lowest energy structures are selected." So as suggested before, I would leave the model where it is.

Hello Shilpa Sarkar , generally after generating a 3D model on servers like I-TASSER and the others (RaptorX, Phyre2 etc). There is need for a refinement of this "crude" 3D structure before validation. However, you may want to assess the how refinement improves on the predicted structure by doing validation before/after refinement. For validation, commonly used tools include, RAMPAGE Ramachandran, ProSA-web server and ProtSAV. Hope this helps.

See, Shey et al., 2019, https://www.nature.com/articles/s41598-019-40833-x

Robert Adamu Shey thank you sir. I have refined the structure using ModRefiner and tried to validate it with SAVES server. The ramachandran plot seems ok with 96% residues in most favourable regions. My only concern is the verify 3D result it shows only 55% residues have score>=0.2 and it should be atleast 80% residues. Any suggestions for further improvement.

Hello Shilpa, sorry for the late response. My suggestion is that may be try refinement on GalaxyWEB, http://galaxy.seoklab.org/ . On the other hand you could try doing 3D prediction with other tools like RaptorX and Phyre2. Then follow up with the refinement and validation.

Dear Robert Adamu Shey . I have tried refining with GalaxyRefine. but the output model has the same problem i.e only around 50% aminoacids have 3d-1d score >=0.2. any suggestion?

Maybe you could try a different approach with the 3D prediction: Phyre 2, Raptor X and Galaxyweb can be of help.

For the validation of protein structure, you can use the "Rampage Server".

• Sujeet Singh
• Could you plz share a working URL for the Ramage server?