On 12:08 Tue 05 May , Anastassis Perrakis wrote:
> Modeller is indeed an excellent program (so are some others) but,
> without being an expert, I was told some time ago and I am still under
> the impression, that homology modeling with 20% identity (and many
> insertions and deletions as you point out) is a dark art. Being able to
> suggest the identical fold with some threading algorithm, is likely all
> you will be able to do. A homology based model might be too much to ask.
Try HHpred. It will do a profile-profile alignment (well, technically
HMM-HMM) and optionally create a multiple-template-based model using
Modeller, all with very little knowledge or input required on the user
end. Evaluating the likely correctness afterwards is a different
story...
--
Thanks,
Donnie
Donnie Berkholz
P. Andrew Karplus lab
Oregon State University
|