I have tried to predicted the 3D structure by using multiple online softwares like I-TASSAR, MUSTER, Swiss-Model and (PS)2 but predicted 3D structure from each software has bad quality and can not be verified by SAVES v5.0. All of these softwares have predicted the structure of only one domain of this protein. I have also done PDB blast to check the structural homologs and i found only 6 homologos and out of these two homologs have similarity with only one domain of my protein.
Kindly guide me what should i do now? I'm working on in-silico nsSNP analysis and i need this structure to check the effect of mutation.
Whether or not you can predict the structure of a protein by homology modeling (Modeller, Swiss-model) or threading (I-Tasser, Muster, PS2) depends on whether or not a related structure can be found in the database of experimental structures, and the degree of sequence homology. For multi-domain proteins you usually get better results if you model each domain individually and try to assemble them later on or perform many analyses on the individual domains.
On individual domains, you can also try ab-initio modeling (template free modeling), however, these programs are usually not available as web servers and are not trivial to use. Look at the results of the last CASP https://en.wikipedia.org/wiki/CASP contest
Article Critical Assessment of Methods of Protein Structure Predicti...
and the CASP website http://predictioncenter.org to see which methods are currently the most successful and what degree of accuracy is possible when they are applied properly.
Be aware of the fact that the lower the homology of your protein to the template, the higher the uncertainty of your model, and therefore the uncertainty of any prediction derived from the model.
- Badges
- Science topic
- Similar topics
- Bioinformatics
- Bioinformatics and Computational Biology