9.6 years ago by
I actually think you need to think more about the science than the tools. There are various modeling tools out there that will produce a model of the structure with your SNP in it. However, the real question is: what are you going to do with that model? I presume that you want to understand whether or not this variation is likely to perturb the structure and/or function of your protein. In my opinion, the data behind the model is going to tell you more about that than the model itself.
My PhD project included some design of mutants to try to understand the role of specific residues in the structure/function of a class of proteins. I built models (using MODELER) of the mutants I was considering, and then I analyzed the data provided by the tool to understand whether the mutant I was considering was likely to introduce a steric clash or other strain. I combined that information with the information from phylogenetic and protein family sequence alignments (produced with CLUSTALW) to form a hypothesis about the likely impact of the mutation. And then I (or someone else in the lab) made the mutations to test the hypotheses.
In my opinion, the algorithms are good, but not really good enough to tell you for certain what the structural perturbations caused by a mutation will be. However, they can give you clues that you can include in your overall analysis.