Deconvolution with OASIS or other methods


Would applying OASIS deconvolution to dF/F traces from the calcium imaging experiments from the Allen Brain Observatory make sense? I would like to know which time points contained spikes.


Hi Maria,

There are already events extracted from the data so you don’t need to do it yourself. You can access it through the SDK this way:

boc = BrainObservatoryCache(manifest_file=manifest_file)
events = boc.get_ophys_experiment_events(ophys_experiment_id=session_id)

where session_id is the ID for the session that you want events for. It will return an array the same shape as what you get for DF/F through the SDK (# neurons X # time stamps).
These events were extracted using the L0 method by Jewell & Witten 2018 (EXACT SPIKE TRAIN INFERENCE VIA ℓ0 OPTIMIZATION - PMC)

You are of course welcome to do your own event detection using other methods - the challenge being you’ll need to find parameters for the event detection that are appropriate for the different Cre lines and indicators (e.g. inhibitory cells have different parameters than excitatory cells).