neural-io-index

name	neural-io-index
description	Index of L0 data-format loader skills by family
layer	L0
group	ephys_wide_band
metadata	{"tags":["index","neural-io","loader","navigation","l0"],"analysis_goal_allowed":["classification","source_localization","feature_extraction","clinical_screening","exploratory","generic","connectivity","phase_amplitude_coupling","online_inference"],"analysis_goal_forbidden":[]}

L0 Neural-IO Format Index

Navigation.
L0 (this page) → L1 Orchestration → L2 Paradigms → L3 Operators.

This index lists L0 IO format loader skills, grouped by file-format family. Each format skill documents the loader library, channel/sample-rate/events mapping into the OperatorIO schema, and known pitfalls. Load one with skill_view(name='<format>') only when _handle_inspect_data hits a non-built-in format.

The dispatcher hits L0 last, after L1 → L2 → L3 — L0 is consulted only when _handle_inspect_data hits a format the built-in dispatcher can't already load; the only legitimate reason to read an L0 skill is "I need to write a custom loader for this format before any pipeline step can proceed."

Format families

Format-skill stubs live as io/<family>/<format>/SKILL.md. The list below matches the planned content drop in T9 sibling plan 03-preprocessing-skill-expansion/3-3-io-formats.md. Anything tagged (planned) below is a slot reserved for that plan; the row should land on this page automatically once the format skill is written.

ephys_wide_band/ — General wide-band electrophysiology (planned: 3)

Format	Extension	Loader
NWB / Neurodata Without Borders	.nwb	pynwb ≥ 2.5
BIDS-iEEG sidecar	sidecar .tsv / .json	mne-bids
MEF3 (Mayo)	.mefd/ directory	pymef

spike_ephys/ — High-rate single-unit ephys (planned: 5)

Format	Extension	Loader
SpikeGLX (Neuropixel)	.bin + .meta	SpikeInterface.read_spikeglx
Open Ephys	continuous.dat + .oebin	SpikeInterface.read_openephys
Blackrock	.ns5 / .ns2 / .nev	brpylib
Plexon	.plx / .pl2	neo.io.PlexonIO
Neuralynx	.ncs / .nse / .nev	neo.io.NeuralynxIO

clinical_ieeg/ — Clinical iEEG / sEEG (planned)

Clinical depth/grid recordings often arrive as proprietary vendor formats plus DICOM-MR for electrode localization. Format skills here cover the loader contract, anonymization expectations, and the BIDS-iEEG export path.

consumer_eeg/ — Mainstream EEG formats (in-code loaders → planned stubs)

Format	Extension	Loader
EDF / EDF+	.edf	mne.io.read_raw_edf
FIF	.fif	mne.io.read_raw_fif
BrainVision	.vhdr + .vmrk + .eeg	mne.io.read_raw_brainvision
EEGLAB SET	.set	mne.io.read_raw_eeglab
LSL XDF	.xdf	pyxdf
CSV / NPY	.csv / .npy	numpy + manual schema
MAT	.mat	scipy.io.loadmat

fnirs/ — fNIRS (planned: 1)

Format	Extension	Loader
SNIRF	.snirf	mne.io.read_raw_snirf

streaming/ — Real-time stream (planned: 1)

Format	Source	Loader
LSL stream	Lab Streaming Layer	pylsl

L1 ↔ L0 integration

When _handle_inspect_data encounters a format that is not in the "in-code loaders" list above (EDF / FIF / BrainVision / EEGLAB SET / XDF / CSV / NPY / MAT), it must:

1. Call skill_view(name='neural-io-index') to discover the right format family.
2. Call skill_view(name='<format>') to fetch the loader contract before writing or generating a custom loader.

See bci/pipeline/SKILL.md Step 1 (Format-Agnostic Loading) for the corresponding orchestration text.

Layering contract

• layer: every format skill carries layer: L0 in frontmatter.
• group: one of {ephys_wide_band, spike_ephys, clinical_ieeg, consumer_eeg, fnirs, streaming}, matching its parent directory (bci/neural-io/<group>/<format>/).
• analysis_goal_allowed/forbidden: optional for L0 (the consistency checker only requires these on L2 paradigm skills). When present, values must come from the 9-enum REGISTRY in easybci_lib/tools/neural_processing/preprocess/analysis_goals.py.

The layer / group enums are defined in easybci_lib/tools/neural_processing/skill_layers.py (single source of truth). Run python -m easybci_lib.tools.neural_processing._check_consistency --strict to verify the contract before merging new format skills.