This capsule uses OASIS to extract neural activity from fluorescence imaging traces through nonnegative deconvolution.
All parameters are passed to run_capsule.py using python run_capsule.py [parameters].
All parameters are defined in main using argparse. The most important one is 'input-dir'
which should point to a directory containing file dff.h5 with the dataset 'data', a 2D array
of processing.json to obtain the frame rate.
The main output is the events_oasis.h5 file.
It contains datasets:
events: The deconvolved neural activity ("events" / "spike rates").
denoised: The inferred denoised fluorescence signal.
If the parameters are automatically estimated, it will also contain the following parameter estimates:
b_hat: The estimated fluorescence baseline value.
lam_hat: The sparsity penalty parameter. Estimated as the optimal Lagrange multiplier for the dual noise constraint problem.
tau_hat: The estimated exponential decay time based on the data's autocovariance.
tau_rise_hat: Optionally, the exponential rise time. Set to zero by default (i.e., negligible), and thus omitted.