I want to get the structure of different proteins. But their similarity is about 95%. There are some mutations distributed in a region of the protein. Which is the best way to obtain a good model, making a in-silico mutation or a homology modelling approach?
95% similarity is quite high, so a homology model would give a pretty good idea. Although it shouldn't take too much time to try both techniques and compare them.
Hi,
95% of total residues, right? may 5 out of 100 or 25 out of 500
So i would prefer homology modeling
What if, the mismatch residues are in contact or in close-proximity.
I'd use homology modelling. It's easy, fast and it'll give you the right answer
I would refer you to check our article for the mutation modeling in case of high similarity. http://www.sciencedirect.com/science/article/pii/S1567134811000992
Making a in-silico mutation will be more helpful, because in homology modeling it may be difficult to identify the difference in the structure due to mutation.
Hi Juan,
I do structure prediction as part of my research so I feel qualified to make the following recommendations.
I would use I-TASSER or ROBETTA to predict the mutant structure. These methods have been among the most accurate methods for structure prediction in the last several CASP competitions where blind predictions needed to be made.
They may take a few days to get your model because of the queues so I also would recommend SPARKS-X which will give you a structure within ~30 minutes and Swiss-MODEL, which also will provide a structure for you quickly. Both methods also perform very well and I have used all 4 successfully in different applications.
Actually for the most scientifically sound prediction, I would predict the structure using all 4 methods and compare the predicted structures to each other. Do they all use the same template? Do they all converge to the same structure (or lets say do 3/4 methods converge to the same structure)? If that's the case, you can with high certainty rely on that structure. Doing in silico mutation may not move the structure enough and since the mutations may cause changes in the structure (i.e. gain or loss of salt-bridges) some of the methods I described above do a refinement step that can capture these effects and movements. Good luck.
Based on the high sequence identity, homology modeling will likely give you your answers very rapidly if the homologous structure is decent quality. The only caveat is if the regions of interest lie in flexible loop regions that can take multiple conformations under different conditions i.e. in the presence / absence of substrates or inhibitors. Then you may have to do mutations and also carry out a wealth of other studies on the mutant.
- Badges
- Science topic
- Similar topics
- Computer Science and Engineering
- Bioinformatics
- Bioinformatic Tools