[gmx-users] Failed to find GROMACS magic number in trr frame header

[Justin Lemkul]

On 4/20/20 6:39 AM, Mijiddorj B wrote:
> Dear Justin,
>
> Thank you very much for your reply. I see. I cut trr file several ns before
> the stopped frame, and I would like to cut the corresponding checkpoint
> file before the stopped frame.

Checkpoint files are not trajectories; they only contain one frame.

> Is it possible to get the specific frame checkpoint using gmx?

Not without using the -cpnum option of mdrun. Checkpoint files are 
overwritten every few minutes so by default, you only ever have the 
current and previously saved states.

> I would like to re-perform the extended simulation again starting from
> another frame. Is it also possible to use the original checkpoint file as
> it is?

Not unless you have a checkpoint file that corresponds to exactly that 
frame, which is usually not the case. Checkpoint files are written every 
few minutes rather than at some fixed interval of time steps.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Assistant Professor
Office: 301 Fralin Hall
Lab: 303 Engel Hall

Virginia Tech Department of Biochemistry
340 West Campus Dr.
Blacksburg, VA 24061

jalemkul at vt.edu | (540) 231-3129
http://www.thelemkullab.com

==================================================