I need to construct a few protein models for which there is no proper template i.e incomplete coverage and even poor identity for the less covered region of the template. How can we construct the model in such a situations.
Towhee, which is a Monte Carlo molecular simulation code, predicts the structure of a protein with a known amino acid sequence.
http://towhee.sourceforge.net/
I'm afraid that only by means of MD or MC with a little hope on success.
Protein modelling without a template is exactly what ab initio methods are for, although you should quickly check first that your target is not (partiall) intrisically disordered (eg disopred, poodle servers). Robetta and Quark servers are places to try ab initio modelling. The latter gives you an estimate of likely accuracy, the former often has big queues. Both servers will use information from any low-identity or incomplete templates that are out there but you are unlikely to get a result that will be as reliable as a normal (>30% sequnce id) homology model, much less an experimental structure. If a proper ab initio model turns out to have a familiar fold, that's a sign that it might be somewhat plausible.
- Badges
- Science topic