AMBER Archive (2008)

Subject: AMBER: problem with nmod

From: yuri pomè (
Date: Fri Mar 28 2008 - 11:42:54 CDT

Dear community,
I'm a college student so I'm new on Amber (and so maybe I'm ignorant
on it as I'm an apprentice :p). I'm working on a several
protein-ligand complex. Trying to calculate entropy with nmod, I've
got a problem with a compound containing a triple bond between two
aliphatic sp1 carbons. During ligand only sander minimization I get an
error due to "LINMIN FAILURE", it sounds strange for a system with
only 34 atoms. So nmod cannot calculate my entropy. The compound's
analogue with double bond, instead triple, works fine (as same as
other analogues).
I've repeated the procedure twice, to check if I did some errors
during parametrization, but I've not figured out nothing! The
procedure is really the same for double-bond analogue and others,
nothing changes.
During minimization the conjugated gradient works bad so the ligand
reach a linmin failure (drms reach 0.1 at most), with steepest-descent
(manually editing ncyc=100000, maxcyc=200000, drms=0,0001) I
can reach a drms of 0,0004 (for example), but when conjugate grad
starts I got this error:

"The system has extended beyond
     the extent of the virtual box.
 Restarting sander will recalculate
    a new virtual box with 30 Angstroms
    extra on each side, if there is a
    restart file for this configuration.
 SANDER BOMB in subroutine Routine: map_coords (ew_force.f)"

Maybe triple bond could bring to a failure during mm_pbsa entropy
calculation? (a very rare bond type in med chem). Or is more probable
that I've done some mistake first? The problem is that I cannot find
any diversity in my procedure between triple and double-bond
analogues, the only difference is the bond type (and so the atom
type!) and that make me crazy! :P
Thanks for all,
Yuri Pomè
The AMBER Mail Reflector
To post, send mail to
To unsubscribe, send "unsubscribe amber" to