Dear AMBER developers and users,
I have encountered an issue while running explicit solvent minimizations with positional restraints using AMBER24, following the standard protocol described in
https://ambermd.org/tutorials/basic/tutorial13/index.php
In my system, I apply positional restraints to all non-solvent atoms during the first minimization step:
restraintmask = ':1-663',
restraint_wt = 250.,
Depending on the system, the number
of residues can vary slightly (e.g., 662 or 663 for my protein alone,
and up to 665–666 when including a Fe-OH cofactor and a substrate).
Here is the problem:
*
Using AMBER24, the job fails silently (i.e., no output, crash before starting minimization cycles) when the mask includes: 1-663 or :1-666, even though these residues exist in the topology.
*
However, the exact same system, inputs, and mask work perfectly in AMBER22.
*
Curiously, AMBER24 only runs the minimization correctly when I use :1-662, even in systems that have 663–666 residues. This is unexpected and inconsistent with the behavior in AMBER22.
The mdout
file in AMBER24 doesn't show any clear error — just the standard
startup information and then stops. Meanwhile, AMBER22 proceeds with the
minimization and produces a full output, even if convergence is not
achieved.
This issue seems to be related to how restraintmask is parsed or applied in AMBER24, possibly in combination with how pmemd is handling restraint groups during minimization.
Has anyone else experienced this? Could this be a regression or stricter parsing introduced in AMBER24?
Any insights or suggestions are
welcome. I’d be happy to provide topology and coordinate files if needed
for reproducing the issue.
Best regards,
Pedro Martínez Zaragoza
PhD Candidate – Computational Chemistry
Universitat Autňnoma de Barcelona
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Fri Jun 06 2025 - 04:00:03 PDT
