aind_ephys_utils.accessors.dataarray module

xarray accessor for exploratory electrophysiology analysis.

This module registers the .ephys accessor for xarray.DataArray.

The accessor is intended to provide a small, composable public surface area that interoperates with standard xarray operations (e.g., .sel, .mean).

class aind_ephys_utils.accessors.dataarray.EphysDataArrayAccessor(xarray_obj: DataArray)

Bases: object

Accessor for exploratory electrophysiology analysis.

Public surface area (intentionally small):

  • da.ephys.validate()

  • da.ephys.align(events, to, window)

  • da.ephys.bin(…)

  • da.ephys.smooth(…)

  • da.ephys.baseline(…)

  • da.ephys.normalize(…)

  • da.ephys.psth(…)

  • da.ephys.reduce(…)

  • da.ephys.restrict(…)

  • da.ephys.plot.<…>

align(events: DataArray | Sequence[float] | ndarray, *, window: Tuple[float, float], to: str | None = None) → DataArray

Align to an event and extract a time window.

Parameters:

  • events – Events object describing event times.

  • to – Event label to align to. Optional for array-like event times.

  • window – (tmin, tmax) window around the event time.

Returns:

Aligned DataArray.

Return type:

xr.DataArray

baseline(*, window: Tuple[float, float], dim: str = 'time', mode: str = 'subtract') → DataArray

Apply baseline correction over a window.

Notes

Applies subtract/divide/zscore baselining over a time window.

bin(dt: float, window: Tuple[float, float] | None = None, output: str = 'rate', time_unit: str = 's') → DataArray

Bin ragged spikes into a dense (trial, unit, time) representation.

Parameters:

  • dt – Bin width in seconds.

  • output – Output type (e.g. “rate” or “count”).

  • time_unit – Unit for time values.

Returns:

Binned spikes.

Return type:

xr.DataArray

Notes

Binning preserves compatible coordinates and updates ephys attrs.

property guide: _Help

Return a short cheatsheet describing the accessor API.

normalize(*, dim: str | Tuple[str, ...], method: str = 'zscore') → DataArray

Normalize data across one or more dimensions.

Notes

Zero-variance slices are set to 0 to avoid NaNs.

property plot: _PlotAccessor

Plot helper sub-accessor.

psth(*, dim: str = 'trial', method: str = 'mean', group_by: str | Sequence[str] | None = None, keep_trials: bool = False) → DataArray

Reduce across trials to compute a PSTH-style summary.

Notes

If keep_trials=True, includes the trial-wise data alongside summary.

reduce(*args: Any, **kwargs: Any) → DataArray

Reduce data dimensionality (e.g. PCA) in an xarray-friendly way.

Notes

Methods include PCA variants, dPCA, and supervised linear reductions.

restrict(*, window: Tuple[float, float], dim: str = 'time') → DataArray

Restrict to a time window for dense or ragged data.

smooth(*, dim: str = 'time', method: str = 'boxcar', sigma: float | None = None, window: float | None = None, boundary: str = 'reflect') → DataArray

Smooth a signal along a dimension.

Notes

Supports gaussian or boxcar smoothing over a single dimension.

validate(*, kind: str | None = None) → DataArray

Validate and return the underlying DataArray.

Parameters:

kind – If provided, require that the DataArray matches the inferred kind.

Returns:

The same object, returned for call chaining.

Return type:

xr.DataArray