cells.spike_triggered_population_coupling_df

cells.spike_triggered_population_coupling_df(
    spikes,
    df_clusters,
    radius_um=500,
    lags=0.08,
    binsize=BINSIZE,
    lowpass_hz=20.0,
    good_units_only=True,
    file_stpc=None,
)

df_clusters / spikes-Bunch convenience wrapper around :func:spike_triggered_population_coupling, for interoperability with the rest of the ephys-atlas cell-features pipeline.

The population that couplings are computed against always includes every cluster in df_clusters regardless of QC – matching Bimbard et al. 2025’s use of “the summed activity of all neurons” as the reference population – irrespective of good_units_only, which only restricts which clusters are reported.

Parameters

Name	Type	Description	Default
spikes	dict - like	Must expose `times` (s) and `clusters` (cluster id, matching `df_clusters`’s index).	required
df_clusters	pandas.DataFrame	Cluster metadata, indexed by cluster id. Must contain `lateral_um`, `axial_um` and `bitwise_fail` (QC flag, 0 == good).	required
radius_um	see :func:`spike_triggered_population_coupling`.		`500`
lags	see :func:`spike_triggered_population_coupling`.		`500`
binsize	see :func:`spike_triggered_population_coupling`.		`500`
lowpass_hz	see :func:`spike_triggered_population_coupling`.		`500`
good_units_only	bool	If True (default, matching this module’s existing cell-features pipeline), only report rows for clusters with `bitwise_fail == 0`. Set to False to report every cluster. `file_stpc`, if given, only ever holds the reported rows, so a cache written with one setting cannot be reused with the other.	`True`
file_stpc	Path	If given and it exists, load `stpc` (already restricted to the reported clusters) from it instead of recomputing; otherwise compute and save it there.	`None`

Returns

Name	Type	Description
	iblutil.util.Bunch	Same as :func:`spike_triggered_population_coupling`, restricted to the reported clusters, plus: df_clusters : pandas.DataFrame `df_clusters` with `coupling_strength` / `coupling_delay` columns added/overwritten (nan for clusters not reported or with no neighbours). i_reported : numpy.ndarray Positional index into `df_clusters` of the rows reported in `stpc` etc.