AllenInstitute / brain_observatory_qc

Other
2 stars 0 forks source link

New Wheel Encoder tag #378

Closed samiamseid closed 6 months ago

samiamseid commented 6 months ago

Metric / Plot Impacted Wheel Encoder and Running Speed

Update Motivation New failure mode discovered for running data. Identified by abnormally flat baseline running speed of 0. Even a stationary mouse should shift slightly and not yield such a flat line. After investigating the movie, it looks like the wheel encoder was too low, causing too much resistance for the wheel to spin freely.

Requested Feature/Solution Add tag called "Excessive wheel resistance: wheel not spinning freely" and have that tag categorized as a fail

Please provide the following information as appropriate:

Screenshots or example plots

wheel_plot_flat

Scope All projects

arielleleon commented 6 months ago

@samiamseid Could you notify engineering? The plot you posted looks problematic and needs to be looked into further. Clark is currently working on the controlled language tag to flag this behavior.

Edit: The plot was because the wheel was bottoming out and I did not catch that in Sam's original post. Engineering does not need to be notified

DowntonCrabby commented 6 months ago

Just updating this ticket with information from the slack thread:

Video for those interested: \allen\programs\mindscope\production\learning\prod0\specimen_1292035016\ophys_session_1300976369\1300976369_Behavior_20231002T090120.mp4

Manual test of the wheel: "\allen\programs\braintv\workgroups\ophysdev\OPhysCore\operator_files\sam_seid\20240105132818_pretest_stim.pkl" Its stored at ['items']['behavior']['encoders'][0]

for some visual context, heres the movie of the test i just did. Wheel starts turning at about 10 seconds "\allen\programs\braintv\workgroups\ophysdev\OPhysCore\operator_files\sam_seid\20240105132818_pretest_Behavior_20240105T132945.mp4"

Here is a plot of the decay speed: decay_speed noiseend cumulative_dist

DowntonCrabby commented 6 months ago

@samiamseid how would you like to distinguish this tag from the already existing tag: Video reviewed: wheel encoder does NOT accurately reflect mouse movement

Are there more specifics you'd like to add to this tag?

samiamseid commented 6 months ago

Yes i think it should be distinguished, because in the case shown above the encoder does exactly match the video. The issue is not with the recording (which would be an instrument failure and require MPE's attention). This issue is with the setup (which is an operators failure, and requires adjustment to the wheel height). It should fail the data not because the data is incorrect, but because the animals movement is impeded

DowntonCrabby commented 6 months ago

Gotcha- it looks like actually we have the incorrect video listed above- but I understand what you are saying now.

Here is a screenshot of the new controlled language tag as part of the wheel encoder & running speed module: image