Errors with "around" selection command in version 0.16

[maxscheurer]

Expected behaviour

Correct number of atoms in selection with "around" keyword.

Actual behaviour

As of version 0.16.0, the atom selection command "around" does not yield the same
results as version 0.15 .
The results gained from the older version are in agreement with VMD's "within" selection keyword, i.e.

set sel [atomselect top "segname A and within 5 of segname B"]

Perhaps I missed something important in the Release Notes that explains this behavior...

Code to reproduce the behaviour

import MDAnalysis as mda

u = mda.Universe(top, trj)
u.select_atoms("segid A and around 5 segid B")

Currently version of MDAnalysis:

0.16.0

[richardjgowers]

This is very possible, I did refactor the around selection. Can you make an example from our test files Max?

…

On Sat, 15 Apr 2017, 6:20 p.m. Max, ***@***.***> wrote: Expected behaviour Correct number of atoms in selection with "around" keyword. Actual behaviour As of version 0.16.0, the atom selection command "around" does not yield the same results as version 0.15 . The results gained from the older version are in agreement with VMD's "within" selection keyword, i.e. set sel [atomselect top "segname A and within 5 of segname B"] Perhaps I missed something important in the Release Notes that explains this behavior... Code to reproduce the behaviour import MDAnalysis as mda u = mda.Universe(top, trj) u.select_atoms("segid A and around 5 segid B") Currently version of MDAnalysis: 0.16.0 — You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHub <#1311>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AI0jB_RVV7BDVadEk3GG5lwNWE-dSId_ks5rwPxigaJpZM4M-aAM> .

[maxscheurer]

I will try to do so asap...

[richardjgowers]

@maxscheurer if it's tricky to make it not work using the test files you can also just upload a system here where you're getting different results

[orbeckst]

Thanks for reporting. An example would be crucial, though.

[orbeckst]

@maxscheurer have you had time to look at an example?

[orbeckst]

ping @maxscheurer ... we would like to look at this issue but we really need something to work with.

[maxscheurer]

@orbeckst, I will try to upload something tomorrow. Thanks for the reminder!

[maxscheurer]

@orbeckst, the issue seems to be fixed in MDAnalysis version 0.16.1
I tested with VMD; mda version 0.15; 0.16.0 and 0.16.1
and the latter one yields the correct result again (as VMD and mda 0.15).

[richardjgowers]

Just in case this crops up later, I think what happened here wasn't actually the around selection itself, but rather the segid A selections working incorrectly in 0.16. In 0.16.1 we improved/fixed how segments were defined in a few cases.