Re: AMBER: Prepin error (?) in xleap

From: David A. Case <>
Date: Thu, 6 Sep 2007 21:38:27 -0700

On Thu, Sep 06, 2007, Lili Peng wrote:
> ( is it always this hard?!).

It's not always this hard, but your case illustrates some of the weaknesses
of our tools. Here are a few comments:

1. Your earlier attempts to load mol2 files failed (in the prmtop creation
step) because they contained atom types unknown to Amber. We should make it
clearer in the documentation that, generally speaking, to really use mol2 as
input format, they should have atom types that the program knows about.

2. The "usual" way to create such mol2 files is to start with a pdb file and
use antechamber. This works in the large majority of cases, but your pdb file
has a bad three-dimensional structure that confuses antechamber. If you could
use Accelrys or something else to get a reasonable starting structure, things
would probably be smoother. Maybe you could do this by hand in xleap or in
whatever modeling program you used to make the GLU3 residue in the first

3. Junmei suggested another workaround, which didn't work for you, perhaps
beacuse of a bug in antechamber. When you run "x = loadmol2 glu3.pdb" you
should see this error:

> x = loadmol2 glu3.mol2
Loading Mol2 file: ./glu3.mol2
Reading MOLECULE named GLU
Fatal Error: last line read: 11 C2 0.1080 0.5590 0.5600
cg 1 0.0000

That should lead you to look at that line in the mol2 file and see the obvious
error (missing "GLU" tags on atoms 10-18). But in doing that, you should also
note that there are no hydrogens in the mol2 file!

At this point, trying to re-make a pdb file is pointless, since you are
missing most of the atoms.

In my view, the best path forward (unless Junmei wants to weigh in with other
ideas) is to try to get a good starting structure in PDB format. You might
consider minimizing with a semiempirical program like mopac, or use other

As I said above, it is certainly a weakness that our tools rely on having a
good starting structure, but there may not be any simple way to remedy that.
[However, a suggestion for Junmei or Wei: we are routinely doing an AM1
geometry optimization to get bcc charges: can't we capture that structure and
make use of it to solve some of the problems like this one?]


The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to
Received on Sun Sep 09 2007 - 06:07:26 PDT
Custom Search