[modeller_usage] Re: Regarding manual protein sequence alignment

[Michal Kurowski <michal AT genesilico.pl>]

Disclaimer: I'm a co-author of one of meta-server methods.

Karsten Suhre [karsten.suhre AT igs.cnrs-mrs.fr] wrote:
> 
> before attacking an alignment by hand, I would try some more sophisticated 
> alignment programs, i.e. those that can include structure information.

This is true and tcoffee is highly recommended but please do not
recommend a single threading-based methods on their own to newcomers.

No single structure prediction method is able to match performance of
combined approach which includes following steps:

- target protein class discovery
- prediction of primary (family membership) and secondary structure
  (including disordered regions)
- structure prediction itself
- approaching the consensus  

I highly appreciate Fugue method as such but what was outlined above
should be kept in mind.

Of course in a simple cases a 'naive' structural alignment, for
example produced by tcoffe in a structural mode might be just in case.

Cheers,

-- 
Michal Kurowski
<michal AT genesilico.pl>