Hi Kankana,
I was following your posts since I have had similar problems before. I am assuming you were following a similar workflow in tutorial39. For similar calculations, I have some walkarounds for my own project, but I’m far from expert with TI calculations, please take them with a grain of salt..
In your original post you said:
The mdout file for eqNTP4 indicates that the simulation did not run, but no error message is provided
I found that in similar cases, pmemd.MPI is more verbose than pmemd.cuda.MPI. If applicable, try running the same simulation with pmemd.MPI may give you some more information to look at. It’s likely something about “velocity exceeding” if you already observed nonphysical contacts/geometry in the generated structure.
I viewed your md input, in which you had:
noshakemask = '.29,46,83,54,78,104,18,19,62,84,14,15'
timask1 = '.1-52'
timask2 = '.53-107'
scmask1 = '.1,20-21,30,40-41,47-52'
scmask2 = '.55,58,63-64,66-67,85-89,92-95'
I usually have the noshakemask cover both timasks, and my timask will cover the respective scmask. For some difficult systems, I extend noshakemask to all solute atoms, and that usually works for me. I have the following suggestions you might play around with your system:
(1) Minimize the transforming region (scmask) as much as possible
If you can run this transformation in separate steps, try to transform fewer atoms in each step.
(2) Make a continuous softcore
I have had some problems with isolated softcores. It might be worth trying to extend the softcore, not skipping the atoms that are directly bonded with or bridging two softcore atoms.
(3) Disable shake for atoms that interact with scmask, to avoid contacts at the SHAKE/no-SHAKE interface
I have had some problems with contacts at the SHAKE/no-SHAKE interface and SHAKE/softcore interface. Please see an archived post from me:
http://archive.ambermd.org/202302/0085.html
I think the contacts at these interfaces may be problematic. Eliminating these types of interfaces helped me get my calculations running. But again, all these assignments may depend on the formalism of the potential and the exact transformations you want to perform.
In your subsequent post, you said that:
But, if I run individual say 0th state and run each step individually
(eqpre1P0, eqpre2P0, eqP0 using pmemd.MPI) then the output file is
correctly written and finished normally.
Sometimes the specific problem occurred only in certain window(s). I had done some transformations where certain windows (lambdas) always failed, so I had to interpolate (if I’m missing only 1 or 2) or redesign the network.
Most of the problems I encountered might be specific to the chemical groups and the biological environment I was trying to simulate. I was hesitant to reply because I’m unsure how much of my experience is transferable to a different system.. I agree with Dr. Simmerling that if you can provide a little more detail about your procedure, the system and the transformation you want to perform, you might be able to get more replies from the community.
Hope this is mildly helpful!
--
Amy He
Forli Lab . TSRI
yihe.scripps.edu
From: Kankana Bhattacharjee via AMBER <amber.ambermd.org>
Date: Wednesday, July 31, 2024 at 1:31 AM
To: AMBER Mailing List <amber.ambermd.org>
Subject: Re: [AMBER] Regarding Advanced Thermodynamic Integration:
But, if I run individual say 0th state and run each step individually
(eqpre1P0, eqpre2P0, eqP0 using pmemd.MPI) then the output file is
correctly written and finished normally. But when I am using through the
code to run the jobs for all windows then it is not able to run the
results. Kindly help me with this (using pmemd.cuda).
Your suggestions would be highly appreciated.
sender notified by
Mail Track for Gmail
<https://urldefense.com/v3/__https://www.mailtrack.email/?utm_source=watermark&utm_medium=email__;!!KGKeukY!yTmvS0Qh2Iu2-ARIKOFuiDmqgT0-rKbBYUVqFK80v-_bgF63PWiaEsizy26bn64vz1Or0I_OwC9fu_w-TeVC5Ui_ZZQ$ >
<https://urldefense.com/v3/__https://www.mailtrack.email/?utm_source=watermark&utm_medium=email__;!!KGKeukY!yTmvS0Qh2Iu2-ARIKOFuiDmqgT0-rKbBYUVqFK80v-_bgF63PWiaEsizy26bn64vz1Or0I_OwC9fu_w-TeVC5Ui_ZZQ$ > [image:
ft34ddz98tt-de616d66631e4648]
On Wed, Jul 31, 2024 at 1:51 AM Kankana Bhattacharjee <
kankana.bhattacharjee_phd21.ashoka.edu.in> wrote:
> Dear AMBER users,
>
> I want to mention that, when performing Advanced TI (Thermodynamic
> Integration) calculation for my protein-ligand system, I am encountering an
> issue where no rst7 file is generated for the eqNTP4 step. The previous
> steps (init, min1, min2, eqpre1P0, eqpre2P0, and eqP0) ran successfully.
> The mdout file for eqNTP4 indicates that the simulation did not run, but
> no error message is provided. Additionally, I had to apply a noshake mask
> for the TI region/atoms due to significant deviations in bond length. I am
> using similar settings for all mdin files as those used for the CDK2
> system in the tutorial.
> I am attaching my eqNTP4.mdin, eqNTP4.mdout (for the two end states).
>
> Your suggestions would be highly appreciated.
>
> Thanks & Regards
> Kankana Bhattacharjee
> Ph.D. Scholar
> Department of Chemistry
> Ashoka University
> Sonipat, Haryana
>
>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
https://urldefense.com/v3/__http://lists.ambermd.org/mailman/listinfo/amber__;!!KGKeukY!yTmvS0Qh2Iu2-ARIKOFuiDmqgT0-rKbBYUVqFK80v-_bgF63PWiaEsizy26bn64vz1Or0I_OwC9fu_w-TeVCyk9wZfE$<https://urldefense.com/v3/__http:/lists.ambermd.org/mailman/listinfo/amber__;!!KGKeukY!yTmvS0Qh2Iu2-ARIKOFuiDmqgT0-rKbBYUVqFK80v-_bgF63PWiaEsizy26bn64vz1Or0I_OwC9fu_w-TeVCyk9wZfE$>
_______________________________________________
AMBER mailing list
AMBER.ambermd.org
http://lists.ambermd.org/mailman/listinfo/amber
Received on Thu Aug 08 2024 - 16:00:02 PDT
