Re: [AMBER-Developers] sqm success/failure on ANP

From: case <>
Date: Thu, 20 Jan 2011 16:27:04 -0500

On Thu, Jan 20, 2011, Ross Walker wrote:

> Just to add my comments to this discussion.

I hope we don't get Eric's nose out of joint here. In fact, he is
providing helpful input by noting that the procedures in the "prepare*"
scripts that dock uses will fail with a common type of ligand. He is
correct, and we ought to be able to do something about it. The short
problem report is that the bcc procedure (as currently implements via
sqm) fails for pyrophosphates. Given that AM1 uses a minimum basis set,
and that it *will* converge when given a good geometry, it seems at least
worth pursuing whether we could not improve SCF convergence for cases like
this. After that, if we thought that the resulting bcc charges were not so
good, we could then turn to improved bond-charge corrections, or really
know that we need to exclude phosphates.

Of course, the larger problems Ross mentioned are still there. My
guess is that many of the ligands that are represented as being highly
negatively charged in the PDB are in fact protonated. Finding out if this
is true is also worth pursuing. And in the end, the best "fix" might turn
out to be to tell people to protonate certain types of phosphates. But this
is all real speculation on my part.

Bottom line: we need to work on this, and not just complain about a naive
request. And we need to thank Eric for bringing this to our attention.

Anyway, that is my two cents....dac

AMBER-Developers mailing list
Received on Thu Jan 20 2011 - 14:00:02 PST
Custom Search