[AMBER] Reproducibility issues for Umbrella Sampling simulations using NFE for different AMBER versions

From: Till El Harrar via AMBER <amber.ambermd.org>
Date: Tue, 21 Jun 2022 18:58:33 +0200

Hello everyone,

I am getting reproducibility issues between AMBER18 and AMBER22 when
using the NFE module, and would like to get some help to understand what
is going on.

I ran Umbrella Sampling simulations on AMBER18 of two amino acids using
the &pmd code with a COM distance restraint as reaction coordinate. The
relative orientation was fixed using two additional COM angle
restraints. When performing the same simulation (identical setup) using
AMBER22, I get vastly more flexible residue movements due to what seems
to be lower restraint strength, resulting in completely different
results. Increasing the restraint forces arbitrarily by 5- to 10-fold
when using AMBER22 results in similar behavior as in AMBER18.

This is most likely caused by an update between AMBER18 and AMBER22, yet
I could not determine the exact change in code leading to the
differences. Could somebody maybe point me to the code origin or to a
knowledgeable person who maintained or updated the code at the time?

Best regards and thanks in advance,


p.s.: I already CCed the authors of the NFE tutorial in.

AMBER mailing list

Received on Thu Aug 04 2022 - 13:32:41 PDT
Custom Search