Monika Sharma wrote: > I am trying to do modelling of a protein, for whom i got the hits for > psi-blast. Now, BLAST showed that the first ~100 residues to be in > homology with one set of proteins, and the rest with other set of > proteins. So for having the desired result, i broke the protein in two > parts, one part of first ~100 residues with two or more proteins showed > by BLAST search, and then the other part with its respectives subjects. > So now i have two modelled parts of protein, but dont know how to link > them, and important thing is that end of the former is the beginning of > the other. My questions are: > 1. Is it the right way of doing? > 2. If yes, can somebody suggest me how to link these parts?

The easiest way to do this is the chimera route, but to get reasonable results you need to know something about the domain-domain interactions. Ideally your two domains should overlap slightly in your alignment so that Modeller has some idea of the relevant distance restraints. Otherwise, probably the best you can do is to perform separate protein-protein docking calculations of the two domains, and use the results to come up with suitable distance restraints for Modeller (e.g. sets of Calpha atoms from the two domains in close proximity).

> 3. An alternative way seems to be like alignment of chimeric protein, > but i am not able to get the desired results in first step of alignment > of the multiple templates. Can there be any way to give your own > alignment files as .ali and .pap? Because when i am trying to give > manually in the usual manner, then the program is looking for the > _fit.pdbs files formed at the first step, and complaining of no match...

It sounds like you have automodel's initial_malign3d turned on. It should definitely be turned off in this case, since it is unlikely that your two templates can be superposed. You can certainly feed in your own alignment file (in PIR format) to automodel - in fact, that's what most of the example scripts do.

Ben Webb, Modeller Caretaker