Cell metrics upgrades

[alexpiet]

• [ ] It would be much faster if we saved out the metrics tables combined for all cells in the platform paper, rather than as individual experiments
• [ ] 'traces' metrics are missing
• [ ] noise_level metric
  • noise_level = np.median(np.abs(np.diff(dff_trace)))/np.sqrt(ophys_frame_rate)
  • https://gcamp6f.com/2021/10/04/large-scale-calcium-imaging-noise-levels/
  • I'm interested in this for the full-session-trace
• [ ] mean_response metric
  • For the full-session-trace.
  • This might just mean np.mean(dff_trace)
• [ ] SNR metric, just the ratio of the mean_response and noise level

[matchings]

question: do we want full trace metrics split by engaged vs. disengaged?

[alexpiet]

Those would probably be nice to have in addition to the full trace not split by engagement. I dont have any immediate use for them.

[alexpiet]

This was solved by https://github.com/AllenInstitute/visual_behavior_analysis/pull/797