AMBER Archive (2004)

Subject: Re: AMBER: Leap atom removal bug?

From: Scott Brozell (
Date: Fri Mar 05 2004 - 20:34:16 CST


Several bugs have been fixed in LEaP and they will be available in
Amber 8.
Here's bugfix 46:

Author: Scott Brozell
Date: 03/05/2004

Programs: LEaP

Description: tleap cause a segmentation fault when removing atoms;
             xleap cause a segmentation fault when Erasing atoms in the
             Unit Editor.

Fix: apply the following patch to amber7/src/leap/src/leap/atom.c

*** atom.c 28 Dec 2001 21:53:15 -0000 7.14
--- atom.c 6 Mar 2004 01:53:17 -0000
*** 908,913 ****
--- 908,914 ----
      for ( i=0; i<iBonds; i++ ) {
          aBond = aaBond[i];
          VP1(( "Breaking bond: %s - %s\n",
+ sAtom1,
                  sContainerFullDescriptor( (CONTAINER)aBond, sAtom2 ) ));
          AtomRemoveBond( aAtom, aBond );

Temporary workarounds: Delete all bonds to an atom before removing it.

On Thu, 4 Mar 2004 wrote:

> Dear all,
> I need to use leap to remove a large number of atoms from a structure, before reading in another PDB file and generating prmtop and prmcrd file. (I'm creating a ligand covalently bound to DNA).
> I would like to do this in a script ( file), but I have found that
> leap crashes with a segmentation fault if I try to remove an atom (using 'remove' text command) when it is bonded to any other atom. If I remove all the relevant bonds ('deletebond') first, so that the atom is isolated, then the 'remove' command works. But identifying all the bonds correctly is a bit tedious, and not easily transferrable to other molecules, which was the point of using a script.
> I guess that this is a bug otherwise there would be no seg fault. Presumably it is the desired behaviour that removing an atom also removes all bonds to it. If anyone can suggest an alternative approach, or is interested in trying to fix the bug, then I'd be very grateful.
> I am using AMBER7 on RedHat 9, compiled with ifort8+gcc3.2.2
> Dave Evans
> London School of Pharmacy
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to