[AMBER] bond equilibrium achieved with sander but not with pmemd, Amber 20

From: Qinghua Liao <scorpio.liao.gmail.com>
Date: Wed, 1 Dec 2021 12:03:31 +0100

Dear Amber developers,

I am simulating a system of a UDP bonded with a glucose, complexed with
a Mn2+ in water.
There is a bond between the UDP and the glucose (O2B -- C1). The
parameter for this bond is 1.46 A, 280 kcal mol-1 A-2.

When I did minimization with pmemd (both CPU and GPU versions), I found
that the bond length was maintained at ~2.1 angstorm,
even though the starting length was around 1.5 angstrom.
(Actually the bond length was kept at 2.1 angstrom for hundreds
nanoseconds of MD without any restraints, until I saw it, never had
this problem before).

However, If I do the minimization with sander, then the bond length can
be successfully kept at ~1.5 angstorm.

With pmemd, I tried to manually modify the bond equilibrium and bond
force constant. If I changed the equilibrium to 3.46 angstrom,
the bond length went to ~3.5 angstrom. If I changed the force constant
to 2800 kcal mol-1 A-2, the bond length went to ~1.8 angstrom.

I have simulated systems of an enzyme in complex with UDP-Glucose-MN2+
using the same parameters, but I don't have this issue.
Simulations were running on the same cluster with the same installation
of Amber 20.

I put the topology, input and minimization files in the Dropbox:
Any comments and suggestions are appreciated.

All the best,

AMBER mailing list
Received on Wed Dec 01 2021 - 03:30:02 PST
Custom Search