AMBER Archive (2005)

Subject: Re: AMBER: crashing xleap; bugfix.44

From: Scott Brozell (sbrozell_at_scripps.edu)
Date: Mon May 23 2005 - 12:22:51 CDT


Hi,

Kristina has diagnosed the problem:

On Fri, 20 May 2005, Furse, Kristina Elisabet wrote:
>
> All of your atoms are listed as type "DU" for dummy atom, so you are getting a
> residue with no atoms in it. When leap tries to create your loop, it cant find
> the atoms to connect, and dies. Just go through and assign atom types and you
> should be fine.

Ilyas has suggested a workaround:

On Fri, 20 May 2005, Ilyas Yildirim wrote:

> I tried to convert it to a .mol2 fine using
>
> # antechamber -i himethyl_jag.prep -fi prepi -o test.mol2 -fo mol
>
> and then load it into xleap with the 'loadmol2' command. It worked this
> way. But I dont know why it gave that segmentation fault.

We have created bugfix.44 to prevent the segmentation fault and improve
the reporting of potential problems:

> loadamberprep himethyl_jag.prep
Loading Prep file: ./himethyl_jag.prep
Entry of type DUmmy; atom C19 not created
Entry of type DUmmy; atom H21 not created
Entry of type DUmmy; atom H22 not created
Entry of type DUmmy; atom H23 not created
Entry of type DUmmy; atom N12 not created
Entry of type DUmmy; atom C13 not created
Entry of type DUmmy; atom H20 not created
Entry of type DUmmy; atom C11 not created
Entry of type DUmmy; atom H18 not created
Entry of type DUmmy; atom N10 not created
Entry of type DUmmy; atom H17 not created
Entry of type DUmmy; atom C09 not created
Entry of type DUmmy; atom C08 not created
Entry of type DUmmy; atom H15 not created
Entry of type DUmmy; atom H16 not created
Entry of type DUmmy; atom C03 not created
Entry of type DUmmy; atom C04 not created
Entry of type DUmmy; atom N05 not created
Entry of type DUmmy; atom C29 not created
Entry of type DUmmy; atom H31 not created
Entry of type DUmmy; atom H32 not created
Entry of type DUmmy; atom H33 not created
Entry of type DUmmy; atom H30 not created
Entry of type DUmmy; atom O06 not created
Entry of type DUmmy; atom H14 not created
Entry of type DUmmy; atom N01 not created
Entry of type DUmmy; atom H02 not created
Entry of type DUmmy; atom C07 not created
Entry of type DUmmy; atom C25 not created
Entry of type DUmmy; atom H26 not created
Entry of type DUmmy; atom H27 not created
Entry of type DUmmy; atom H28 not created
Entry of type DUmmy; atom O24 not created
** LOOP atom C09 not found - bond not formed
Discarding residue (This is a remark line) to EOF

********>Bugfix 44:
Author: Scott Brozell
Date: 05/20/2005

Programs: LEaP

Description: tleap and xleap may cause a segmentation fault when loading
             a prep file containing LOOP bonds between atoms of type dummy.

Fix: apply the following patch to amber8/src/leap/src/leap/amber.c

------------------------------------------------------------------------------
*** amber.c 29 Apr 2002 16:48:37 -0000 7.17
--- amber.c 21 May 2005 06:51:33 -0000
***************
*** 1189,1194 ****
--- 1189,1199 ----
                  aMain1 = aAtom;
              }
          }
+ else {
+ if ( strncmp( sName, "DUMM", 4 ) != 0 ) {
+ VP0(( "Entry of type DUmmy; atom %s not created\n", sName ));
+ }
+ }
          cPCoor->aAtom = aAtom;
          iIndex++;
      }
***************
*** 1249,1260 ****
                  aAtom2= (ATOM)cContainerFindName( (CONTAINER)rRes,
                                                        ATOMid, saStr[2] );
                if ( aAtom == NULL ) {
! VP0(( "** LOOP atom %s not found\n", saStr[1] ));
                    problem++;
                }
                if ( aAtom2 == NULL ) {
! VP0(( "** LOOP atom %s not found\n", saStr[2] ));
                    problem++;
                }
                if (bAtomBondedTo(aAtom, aAtom2)) {
                    VP0(( "%s: LOOP: redundant bond %s--%s ignored\n",
--- 1254,1269 ----
                  aAtom2= (ATOM)cContainerFindName( (CONTAINER)rRes,
                                                        ATOMid, saStr[2] );
                if ( aAtom == NULL ) {
! VP0(( "** LOOP atom %s not found - bond not formed\n",
! saStr[1] ));
                    problem++;
+ continue;
                }
                if ( aAtom2 == NULL ) {
! VP0(( "** LOOP atom %s not found - bond not formed\n",
! saStr[2] ));
                    problem++;
+ continue;
                }
                if (bAtomBondedTo(aAtom, aAtom2)) {
                    VP0(( "%s: LOOP: redundant bond %s--%s ignored\n",
------------------------------------------------------------------------------

Workarounds: Verify the LOOP bonds. Correct the dummy types of the non-dummy
             atoms; this may be accomplished in several ways. One way is
             by using a file format other than the prep file format.
             For example, the mol2 file format is apt for a single residue.
             antechamber can convert between formats, e.g.:
             antechamber -i input_file -fi prepi -o output_file -fo mol2

> Quoting Kenley Barrett <kenley.barrett_at_gmail.com>:
> >
> > Thanks for your help. I have attached the prep file. I am using Amber
> > 8, and I open XLeap using the command:
> >
> > /opt/local/amber8/exe/xleap -s -f /opt/local/amber8/dat/leap/cmd/leaprc.ff99
> >
> > The first command that I then enter into XLeap is:
> >
> > loadamberprep himethyl_jag.prep
> >
> > The program immediately crashes.
> >
> > On 5/20/05, Scott Brozell <sbrozell_at_scripps.edu> wrote:
> > > Hi Kenley,
> > >
> > > I am not aware of this bug. If you send me your input files, your
> > > leap.log for the commands you attempted, and some details like which
> > > amber version, what computing platform, etc then I'll try to
> > > reproduce your bug.
> > >
> > > Scott Brozell
> > >
> > > On Fri, 20 May 2005, Kenley Barrett wrote:
> > >
> > > > Amber users,
> > > >
> > > > I am trying to create a library file for an unusual residue. I have
> > > > created a prep file, but whenever I try the XLeap command
> > > > "loadamberprep [jobname].prep", XLeap immediately crashes due to a
> > > > "segmentation fault". Has anyone else had this problem, and does
> > > > anyone know how to fix it?
-----------------------------------------------------------------------
The AMBER Mail Reflector
To post, send mail to amber_at_scripps.edu
To unsubscribe, send "unsubscribe amber" to majordomo_at_scripps.edu