Conventions¶
This package relies on stable naming conventions for dimensions, coordinates,
and attributes. If you use the from_dataframe method to build your analysis object, it will be formatted correctly by default. If you create an xarray.DataArray object yourself, it should contain the following information:
Dimensions¶
trial: trial index for trial-aligned dataunit: neuron/unit indextime: time samples in seconds (or other unit inephys.time_unit)component: unit axis may be replaced by component axis after dimensionality reduction
Attributes¶
ephys.kind:"spikes_ragged","binned", or"continuous"ephys.timebase:"session"or"trial"ephys.time_unit: typically"s"ephys.valid_intervals: list of(tmin, tmax)tuplesephys.history: list of processing steps (optional)
Defaults for adapters¶
When using from_dataframe, the default column names are:
spike_times: spike times in the units tablestart_time: trial start time in the trials tableend_time(orstop_time): trial end time in the trials table
Events representation¶
The adapter builds events as a DataArray with dims (trial, event, bound),
where bound contains "start" and "end". The align op accepts
this representation directly and will use the "start" bound by default.
Where to find definitions¶
All constants are defined in aind_ephys_utils.standards.conventions.