ADL: [ADL] Autodock Vina - Cluster Version?

Eike Schulz eschulz1 at gwdg.de
Sat Jun 20 13:19:17 PDT 2009


Hello everybody,

I am >very< new to docking simulations and happy that Autodock-Vina is so
simple to use, and indeed it does result in rigid-body docks in a very short
time.

However, I have some questions and this far I could not find any resource to
look for the answer.

1. It is my feeling that the search space should be of comparable size to
the ligand? But does it have to be in every direction?
2. I am working with a polymeric ligand of nearly 5kDa that extends to size
of ~70Å ­ as you can guess the search space is becoming rather large and
computation does take rather long.
3. Is there a version of Autodock-Vina that is compatible with a
ŒBeowulf¹-kind cluster. We have combined 21 DualCore Systems (2 cores + HT)
with MosX. It would be great if one could run Vina on 80 cores instead of 8.
Maybe there is beta version or a version that could be compiled to work on
such a cluster? 
4. If Autodock-Vina is not capable to do that ­ what are your suggestions to
try instead?

After those technical questions I have some nasty general questions:

How can you tell how reliable a solution is? For some of my dockings I can
tell from homology to crystal structures. But what if there is nothing you
can compare it to? How much do calculations benefit from flexible
sidechains? I reckon that it improves Œreliability¹ a lot but the cost is
computation time. So how do you balance it? The optimal case would certainly
be to leave all sidechains flexible, but that takes forever. Next aproach is
to leave only those in a binding area flexible ­ but sometimes that affected
the results unexpectedly ...

I hope those questions are not too general or novice-like for such a mailing
list.


Kind regards.

Eike


More information about the autodock mailing list