Problem with "Data range too large for int16"

[Lingjian Wang]

I'm new with the MSMbuilder, while I followed the tutorial of alanine dipeptide without any problem, I encounter the problem of "Data range too large for int 16" with my own testing trajectory .

My input file is a 20ns chymotrypsin(with approximately 3000 atoms for dry protein ) run using AMBER, which have been converted into XTC format using VMD.

I put the trajectory file into the following path: ./test/TRAJ00/frame0.xtc
The corresponding PDB file is in the root directory.

Then I run the following command in the terminal:
sudo ConvertDataToHDF.py -s test.pdb -i test

It gives me an error like this:
"Data range too large for int16: try removing center of mass motion, check for 'blowing up, or just use .h5 or .xtc format.'

Could anyone tell me what is the problem and how can I fix it? Is there any specific rules for the input file that I don't know?

[Robert McGibbon]

MSMBuilder saves your trajectories in a fixed point representation, so it's warning you that the xyz coordinates are too big to be handled without errors.

The most obvious cause of this would be if your XTC is not centered. Moving the center of mass to the origin would substantially reduce the magnitude of the xyz coordinates.

[Lingjian Wang]

Thanks,Robert. I centered my AMBER trajectories and now I can continue with MSM analysis without this data range issue.
But I wonder why use int16 but not int32 or int64? I'm not familiar with programing,maybe for the purpose of saving RAM space?