Problem with select_atoms

[katieesconfused]

Dear MDAnalysis team and Users,

I'm trying use MDAnalysis select_atoms function to dump out atom index, frame and positions of species at the Interface of a system.
Problem : the output is WRONG, and very specifically wrong.

This is my code:
import MDAnalysis as mda
import numpy as np

topo = "/scratch/anhnguyen99/g/s/si2/nvt/p10/ps.gro"
trjnj = "/scratch/anhnguyen99/g/s/si2/nvt/p10/2nd10ns/ps2nj.xtc"
unj = mda.Universe(topo, trjnj, format='xtc',topology_format="gro")
unj.trajectory[500]

linj = unj.select_atoms("name LI and (around 5 resname PE120)", periodic = False, updating= True)
print(linj.indices)

Result:
[44338 44355 44405 44436 44446]

For demonstration I'm slicing at frame 500 in the trajectory.
The trajectory is converted to remove box periodicity using gromacs gmx trjconv -pbc nojump,
This problem persist even when I use the original wrapped trajectory.

When I visualize all of species that got selected this way, this is the result, which is INCORRECT:

When I use GROMACS gmx select, I get these indices as results:
44339 44406 44420 44441 44443 44445 44447

and this is what it looks like, which is CORRECT, for my purpose:

The atoms that got selected by MDAnalysis , for some of them, the index is off by exactly 1
44338 instead of 44339
44405 instead of 44406
44446 instead of 44447

I already set period = False for geometric selection, and updating = True for dynamic selection in the trajectory.
MDAnalysis version I'm using is MDAnalysis 2.7.0.

Any help is greatly appreciated!! If you have some time to spare please help and take a look at this, I suspect there is a bug with the function select_atoms of MDAnalysis 2.7.0.

Thank you for your time, please let me know what other information I can provide to on this issue.
Anh.

[hmacdope]

@katieesconfused

A few issues,

• you will need to set periodic=True when selecting with MDAnalysis otherwise some distances over the PBC will not be taken into account.
• How are you visualizing? This is very important as could lead to a mismatch in the atoms selected. Beware of any "off by one" bugs when comparing indices. MDAnalysis indicies are 0 indexed, (like python). I am not sure what the indexing is like in your visualization program.