On 12/27/14 7:35 PM, Adhikari, Badri (MU-Student) wrote:
I used Modeller to reconstruct around a 100 proteins with true
residue-residue contacts and secondary structure information. I could
reconstruct only a few proteins with high accuracy. Below is a sample
script I used for Modelling. I did not use any templates. (I found that
Modeller satisfies more contact restraints when I provided a template
with just one residue’s coordinates instead of an extended structure as
a template.) I wonder why Modeller could not satisfy true contact
restraints. Is there any specific way to reconstruct proteins using true
residue contacts?
You need a lot of these contacts to model a protein. Look at the
restraint file from a typical Modeller run - it includes a large number
of atom-atom distances with a wide span of residue ranges. I'm assuming
you have more than 4 CA/CB contacts, but you probably don't have enough.
In addition, secondary structure restraints are pretty weak (they're
more of a "suggestion") and are unlikely to force the protein to adopt
that secondary structure unless it's pretty close to start with.