This capsule is designed to curate spike sorted data for the AIND pipeline.
It uses a quality metrics based recipe to flag units as passing or failing default quality control (QC).
The recipe is based on the following quality metrics:
- isi violation ratio < 0.5
- presence ratio > 0.8
- amplitude cutoff < 0.1
The data/
folder must include the output of the aind-ephys-postprocessing, including the postprocessed_{recording_name}
folder.
The code/run
script takes no arguments.
A list of curation thresholds used for curation can be found in the code/params.json
:
{
"job_kwargs": {
"chunk_duration": "1s",
"progress_bar": false
},
"curation": {
"isi_violations_ratio_threshold": 0.5,
"presence_ratio_threshold": 0.8,
"amplitude_cutoff_threshold": 0.1
}
}
The output of this capsule is the following:
results/qc_{recording_name}.npy
file, containing the boolean labels from the automatic curationresults/data_process_curation_{recording_name}.json
file, a JSON file containing aDataProcess
object from the aind-data-schema package.