You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hi!
The aind-data-schema is currently not a dependency of this project and i would like to keep it that way since anyone can use these definitions to derive tasks without use the aind-data-schemas-(models). That being said, I do acknowledge that at some point we need to interact with the existing infrastructure. To achieve that I have been pushing all that code into https://github.com/AllenNeuralDynamics/Aind.Behavior.ExperimentLauncher.
The modality/watchdog integration, in particular, follows the strategy we discussed previously regarding waterlog-slims by using a aind session schema to instantiate the watchdog manifest (since the schema must know what modalities were used during the session). This means that the only point where Modality must be known is really at the session schema creation. I think this is a fair ask since, currently, in order to upload via watchdog, users are expected to provide a valid session.json anyway.
How to create the session.json is up to the user since it will be largely experiment-specific (as I am sure you are already doing too). For instance, I run this light weight mapper at the end of each session that determines whether video data was acquired or not and appends the modality.
You could use a similar strategy to check if FIB, Ephys, etc... was also present but checking either the expected data output or even the rig schema instance that spawned the session.
If you feel its necessary to have it at the session level we can talk about wrapping the class that is needed since issues like AllenNeuralDynamics/aind-data-schema#960 (comment) make me a bit nervous adding aind-data-schema and the Modality class as a dependency to this package.
The aind watchdog service has a concept of modality: what data types are being recorded for example behavior, video, fiber photometry, ephys, etc.
Does this information belong in the session model?
The text was updated successfully, but these errors were encountered: