We have been reading session type and date of acquisition from pkl files, due to issues with the LIMS database.
The behavior_sessions table in LIMS lacks stimulus_name, we used to read this from mtrain, but mtrain contained incorrect values so we read them from the pkl files instead. This doesn't scale.
the date_of_acquisition value in the pkl file should be groundtruth. There is date_of_acquisition in behavior_sessions, ophys_sessions, and ecephys_sessions table, and they might differ. Compare against pkl file to see which table contains errors, and work with LIMS team to fix those.
Creating this ticket to track progress on this. Work for this will most likely fall on LIMS team.
Once LIMS db is fixed, then update allensdk to read from there instead of pkl.
We have been reading session type and date of acquisition from pkl files, due to issues with the LIMS database.
The behavior_sessions table in LIMS lacks stimulus_name, we used to read this from mtrain, but mtrain contained incorrect values so we read them from the pkl files instead. This doesn't scale.
the date_of_acquisition value in the pkl file should be groundtruth. There is date_of_acquisition in behavior_sessions, ophys_sessions, and ecephys_sessions table, and they might differ. Compare against pkl file to see which table contains errors, and work with LIMS team to fix those.
Creating this ticket to track progress on this. Work for this will most likely fall on LIMS team.
Once LIMS db is fixed, then update allensdk to read from there instead of pkl.