I am using flexible enforced rotation on a long narrow protein against the entrance to an enzyme active site. I am trying to rotate the narrow protein 360 degrees so I can then capture windows for umbrella sampling. 

If I used iso-pivot-free rotation the protein becomes distorted to the point that I end up with LINCS errors. I found that this was protein length dependent. A smaller prototype protein rotated nicely, but with the full human protein I encountered problems. So I am now trying out flexible rotation over a range of rot-rate0 and rot-k0 values. 

I have a total of 12 jobs in the queue. I've noticed the minute that one crashes, the next starts. But I do wonder whether the next starting is causing the original to crash! I've never tried using flexible rotation, I notice that it produces a number of additional files on top of the usual .log file from Gromacs. 

Simulation 1 got to 1.250 ns before crashing. 

Simulation 2 got to 1.600 ns before crashing. 

The only difference is rot-k0 was 600 in Simulation 1 and 500 in Simulation 2. 

The exact errors I am getting in both cases are:

"Fatal error:

Enforced rotation: No reference data for first slab (n=-14), unable to proceed."

My concern is that it can no longer find reference data during an arbitrary point in the simulation, hence why I am wondering whether there an IO problem between job. However, searching files on time stamp doesn't reveal anything.

I have also noticed that the gromacs log file has the following line:

"Enforced rotation: allocating memory to store data for 29 slabs (rotation group 0)"

Perhaps this could be a memory issue?

.mdp file details are below. 

rotation                 = Yes

rot-nstrout              = 1

rot-ngroups              = 1

rot-group0               = Collagen_CA

rot-type0                = flex 

rot-massw0               = yes

rot-vec0 = 0.691585601358247 -0.6995947474399045 0.17965674311989613

rot-rate0                = 0.021  

rot-k0                   = 100.0

rot-slab-dist0           = 1

rot-min-gauss0           = 0.001

rot-eps0                 = 0.0001

rot-fit-method0          = norm

rot-potfit-nsteps0       = 21

rot-potfit-step0         = 0.25

Thanks

Anthony