Closed dougollerenshaw closed 3 years ago
@dougollerenshaw This is useful thing to cache.
I have a similar table already computed at /nc-ophys/visual_behavior/behavior_model_output/_summary_table.pkl
It might be good to think about whether we should combine these tables. My summary table as behavioral model outputs, but also model-free statistics like when the mouse was engaged, the lick rate, the reward rate, number of hits, number of engaged hits, etc.
Thanks @alexpiet. Your table is only for ophys sessions, correct? When I load it, it's only 346 entries long: As paper figure 1 is currently outlined, it calls for summarizing behavior performance across all behavior sessions. While I'm not sure it makes sense to show all stages, I think it might make sense to at least show the last training day and/or the last pre-imaging habituation day. So we'll need the outputs of my method to get those sessions.
Thats correct that it is only for ophys sessions. There is a second table that has a subset of columns for all behavioral sessions _training_summary_table.csv
. I've been meaning to update that table to include all the same values as the ophys table, but haven't updated it yet
If its too much work to think about how to combine these tables, its fine. I just wanted you to be aware that my tables existed.
To summarize behavior data across sessions, I needed a more convenient/fast method for getting access to behavior data for all sessions. I opted to add some infrastructure to pre-compute behavior performance for all sessions using the cluster, which is included in this PR. We will then be able to quickly open the cached data for sessions of interest and compute meta-summary data
This PR: