README.md

Curation for AIND ephys pipeline

aind-ephys-curation

Description

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

Inputs

The data/ folder must include the output of the aind-ephys-postprocessing, including the postprocessed_{recording_name} folder.

Parameters

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
    }
}

Output

The output of this capsule is the following:

• results/qc_{recording_name}.npy file, containing the boolean labels from the automatic curation
• results/data_process_curation_{recording_name}.json file, a JSON file containing a DataProcess object from the aind-data-schema package.