Chaix Denis wrote: > I want to use have Modeller, DNA-protein complexes. I have a structure > of a protein:DNA as a template and the sequence of an unknown dna > binding protein structure > I have already made the alignement > Could modeller model the unknown protein and the unknown DNA structure? > What is the good protocol to obtain a good model

Modeller cannot do "homology modeling" of DNA, so if your DNA sequence in the target is different from the template, you're out of luck. But if the sequence is the same, you can treat the DNA like any other ligand and just copy it rigidly from the template to your model, using the '.' residue type in your alignment.

Of course Modeller can model your protein, but it has no explicit protein-DNA binding potential - it'll just preserve in your model any close interactions between protein and DNA that were in the template. This should be OK if you have reasonable sequence identity in the binding region between target and template.

Ben Webb, Modeller Caretaker