radiens_core.models.spikesorter

Spike sorter control and status domain models.

CLASS	DESCRIPTION
DashElement	Elements available for the spike sorter dashboard.
InputFeatureType	Feature types used for spike sorting.
SpikeDetectParams	Detection parameters for a single recording channel.
SpikeSorterClustererParams	Parameters for the spike clustering algorithm.
SpikeSorterCommand	Commands for controlling an active spike sorter.
SpikeSorterCriterionLevel	Stringency levels for AI evaluation, from least to most strict.
SpikeSorterDashboard	Comprehensive spike sorter dashboard data.
SpikeSorterDashboardGeneral	General summary metrics for the spike sorter dashboard.
SpikeSorterDynamicConfig	Configuration for AI dynamic tracking auto-updates.
SpikeSorterDynamicCriteria	Criteria stringency levels for AI auto-updates.
SpikeSorterDynamicUpdatePeriod	Update periods in seconds for AI auto-updates.
SpikeSorterFeatureParams	Parameters for feature extraction during spike sorting.
SpikeSorterLaunchParams	Parameters for launching a new spike sorter.
SpikeSorterLaunchResult	Result of a spike sorter launch request.
SpikeSorterParamCommand	Commands for updating individual spike sorter detection parameters.
SpikeSorterPhase	Where an offline sort is in its pipeline.
SpikeSorterState	Current state and progress of a spike sorter.
SpikeSorterStateSystem	Overall system state of a spike sorter.
SpikeSorterSubCommand	Sub-commands for spike sorter operations.
SpikeSorterType	Supported spike sorter types.

ATTRIBUTE	DESCRIPTION
FlexDashElement	DashElement that also accepts a string name (case-insensitive) or int value.
FlexSpikeSorterCommand	SpikeSorterCommand that also accepts a string name (case-insensitive).
FlexSpikeSorterSubCommand	SpikeSorterSubCommand that also accepts a string name (case-insensitive).

Attributes

FlexDashElement module-attribute

FlexDashElement = DashElement | DashElementStr | int

DashElement that also accepts a string name (case-insensitive) or int value.

FlexSpikeSorterCommand module-attribute

FlexSpikeSorterCommand = SpikeSorterCommand | SpikeSorterCommandStr

SpikeSorterCommand that also accepts a string name (case-insensitive).

FlexSpikeSorterSubCommand module-attribute

FlexSpikeSorterSubCommand = SpikeSorterSubCommand | SpikeSorterSubCommandStr

SpikeSorterSubCommand that also accepts a string name (case-insensitive).

Classes

DashElement

Bases: IntEnum

Elements available for the spike sorter dashboard.

ATTRIBUTE	DESCRIPTION
ENABLED_PORTS	Ports the sorter is enabled on.
GENERAL	Sorter-wide summary.
PORT_STATS	Per-port statistics.
SITE_STATS	Per-site statistics.
SITE_PANEL	Per-site panel data.
NEURON_STATS	Per-neuron statistics.
NEURON_PANEL	Per-neuron panel data.
AI	AI tracking state.

InputFeatureType

Bases: IntEnum

Feature types used for spike sorting.

SPD and SPD_POS are not supported at the moment.

ATTRIBUTE	DESCRIPTION
WFM	Spike waveform.
SPD	Shape, phase and distribution features.
WFM_POS	Waveform plus site position.
SPD_POS	Shape, phase and distribution features plus site position.

SpikeDetectParams

Bases: BaseModel

Detection parameters for a single recording channel.

All fields are optional to allow for partial updates.

ATTRIBUTE	DESCRIPTION
is_enabled	Whether detection is enabled on this channel. TYPE: bool | None
ntv_chan_idx	Native channel index (used in return structures, not set via bulk update). TYPE: int | None
thr	Threshold values in microvolts [neg, pos]. TYPE: list[float] | None
thr_sd	Threshold values in standard deviations [neg, pos]. TYPE: list[float] | None
thr_wdw_pts	thr_wdw in samples. Read-only; update_detect_params() ignores it. TYPE: list[int] | None
thr_wdw	Spike window around the threshold crossing in seconds [pre, post, total]. update_detect_params() sends only pre and post; the server reports the window with its peak-search margin added. TYPE: list[float] | None
shadow	Dead time after a detected spike during which no new spike is detected, in seconds. TYPE: float | None
shadow_pts	shadow in samples. Read-only; update_detect_params() ignores it. TYPE: int | None
is_set_thr	Whether the detection threshold is armed [neg, pos]. TYPE: list[bool] | None
weak_thr	Artifact-rejection level in microvolts [neg, pos]. When armed, a detected waveform that goes past it is discarded as an artifact. TYPE: list[float] | None
weak_thr_sd	Artifact-rejection level in noise standard deviations [neg, pos]. TYPE: list[float] | None
is_set_weak_thr	Whether the artifact-rejection level is armed [neg, pos]. Setting a level does not arm it. TYPE: list[bool] | None
spike_width_sec	[min, max] seconds a detected spike may span at half its peak on its detecting site and still be clustered. One outside the bounds is marked noise and never assigned to a unit. Defaults to [0.00015, 0.002]; the 0.15 ms minimum rejects narrow spikes such as cerebellar ones (about 0.1 ms). TYPE: list[float] | None

SpikeSorterClustererParams

Bases: BaseModel

Parameters for the spike clustering algorithm.

ATTRIBUTE	DESCRIPTION
pos_influence_frac	Influence of spike position on clustering. TYPE: float | None
wfm_shape_influence_frac	Influence of waveform shape on clustering. TYPE: float | None
eps	Epsilon parameter for DBSCAN-like clustering. TYPE: float | None
min_pts	Minimum points for a cluster. TYPE: int | None
disabled	Whether clustering is disabled. TYPE: bool

SpikeSorterCommand

Bases: IntEnum

Commands for controlling an active spike sorter.

ATTRIBUTE	DESCRIPTION
ON	Start sorting.
OFF	Stop sorting.
CLEAR_SORT	Discard the sorter's spikes.
REBASE	Stop, discard the spikes, apply new feature and clusterer parameters, and start again.
INIT	Stop, discard the spikes, reset detection parameters to defaults and rebuild the site neighborhoods. A sorter that was running is started again.

SpikeSorterCriterionLevel

Bases: IntEnum

Stringency levels for AI evaluation, from least to most strict.

ATTRIBUTE	DESCRIPTION
LEAST	Least strict.
LESS	Less strict than balanced.
BALANCED	The middle level.
MORE	More strict than balanced.
MOST	Most strict.

SpikeSorterDashboard

Bases: BaseModel

Comprehensive spike sorter dashboard data.

ATTRIBUTE	DESCRIPTION
enabled_ports	List of hardware ports with active sorting. TYPE: list[Port]
general	General summary metrics. TYPE: SpikeSorterDashboardGeneral | None

SpikeSorterDashboardGeneral

Bases: BaseModel

General summary metrics for the spike sorter dashboard.

SpikeSorterDynamicConfig

Bases: BaseModel

Configuration for AI dynamic tracking auto-updates.

ATTRIBUTE	DESCRIPTION
is_enabled	Whether dynamic AI tracking is enabled. TYPE: bool | None
update_period_sec	Schedule for dynamic updates. TYPE: SpikeSorterDynamicUpdatePeriod | None
criteria	Stringency criteria for updates. TYPE: SpikeSorterDynamicCriteria | None

SpikeSorterDynamicCriteria

Bases: BaseModel

Criteria stringency levels for AI auto-updates.

SpikeSorterDynamicUpdatePeriod

Bases: BaseModel

Update periods in seconds for AI auto-updates.

SpikeSorterFeatureParams

Bases: BaseModel

Parameters for feature extraction during spike sorting.

ATTRIBUTE	DESCRIPTION
num_features	Number of features to extract. TYPE: int
feature_type	Type of features to extract: WFM or WFM_POS. Also accepts the name as a string (e.g. "wfm"). Defaults to WFM_POS, matching the server. TYPE: InputFeatureType | InputFeatureTypeStr
latent_dim	Latent dimension for feature reduction. TYPE: int | None
resample_wfm_len	Length to resample waveforms to. TYPE: int | None

SpikeSorterLaunchParams

Bases: BaseModel

Parameters for launching a new spike sorter.

ATTRIBUTE	DESCRIPTION
dsource_id	ID of the data source to sort (required for file sorting). TYPE: str | None
sub_cmd	Sub-command for the sorting operation. SpikeSorterSubCommand.DETECT_ONLY detects spikes without clustering them into neurons. TYPE: SpikeSorterSubCommand
clusterer_params	Optional clustering parameters. None uses the server's defaults, as does any field left None. With an older Radiens server, None keeps whatever an earlier launch on the dataset set. TYPE: SpikeSorterClustererParams | None
feature_params	Optional feature extraction parameters, with the same None behavior. TYPE: SpikeSorterFeatureParams | None
neighbor_radius_um	Per-site neighbor search radius, in microns. None keeps the dataset's current radius, which an earlier launch may have changed. TYPE: float | None
max_num_site_neighbors	Cap on the number of neighbors kept per site (closest first). 0 detects on every channel independently, so a threshold crossing is never discarded for being smaller than a neighbor's. None keeps all neighbors within the radius when neighbor_radius_um is set; with both None the current neighborhood is kept. TYPE: int | None
seed	Root of every random draw the sort makes, recorded in the sort's provenance. None draws one from the clock. Passing the seed a previous sort recorded reproduces that sort, given the same recording, settings and server build. Detect-only sorts record no seed. TYPE: int | None

SpikeSorterLaunchResult

Bases: BaseModel

Result of a spike sorter launch request.

ATTRIBUTE	DESCRIPTION
spike_sorter_id	Unique ID of the launched sorter. TYPE: str
sorter_type	Type of the launched sorter. TYPE: SpikeSorterType
msg	Status message from the server. TYPE: str

SpikeSorterParamCommand

Bases: IntEnum

Commands for updating individual spike sorter detection parameters.

update_detect_params() sends these for you, one per SpikeDetectParams field set.

ATTRIBUTE	DESCRIPTION
THR_LEVEL	Sets thr.
THR_LEVEL_SD	Sets thr_sd.
THR_WDW	Sets thr_wdw.
WEAK_THR_LEVEL	Sets weak_thr.
WEAK_THR_LEVEL_SD	Sets weak_thr_sd.
SHADOW	Sets shadow.
THR_ACTIVATE	Sets is_set_thr.
WEAK_THR_ACTIVATE	Sets is_set_weak_thr.
SPIKE_WIDTH	Sets spike_width_sec.

SpikeSorterPhase

Bases: IntEnum

Where an offline sort is in its pipeline.

A streaming sorter's phase has no fixed meaning. For a recording still being written, DETECTING and FINALIZING can last until recording stops. Older servers always report UNSPECIFIED.

ATTRIBUTE	DESCRIPTION
UNSPECIFIED	The sort has not started.
DETECTING	Detecting spikes.
TRAINING	Training the sorter on the detected spikes.
SORTING	Clustering spikes into units.
FINALIZING	Writing out the result.
COMPLETE	Done. The output fileset can be read.

SpikeSorterState

Bases: BaseModel

Current state and progress of a spike sorter.

ATTRIBUTE	DESCRIPTION
sys	Overall system state (ON, OFF, NOT_CONFIGURED). TYPE: SpikeSorterStateSystem
phase	Which stage of the pipeline an offline sort is in. TYPE: SpikeSorterPhase
frac_complete	Fraction of the current phase completed (0.0 to 1.0). It returns to 0 at each phase change and reads 1 at the end of training as well as at the end of the sort, so a caller waiting on a sort reads phase, not this. TYPE: float
is_on	Whether the sorter is currently active. TYPE: bool
msg	phase as a display string. TYPE: str
error_msg	Error message if the sorter is in an error state. TYPE: str
warn_msg	Warning message from the server. TYPE: str

SpikeSorterStateSystem

Bases: IntEnum

Overall system state of a spike sorter.

ATTRIBUTE	DESCRIPTION
ON	Running.
OFF	Stopped.
NOT_CONFIGURED	Not set up yet.

SpikeSorterSubCommand

Bases: IntEnum

Sub-commands for spike sorter operations.

ATTRIBUTE	DESCRIPTION
NULL	Full sorting, with clustering.
DETECT_ONLY	Detect spikes without clustering them.

SpikeSorterType

Bases: IntEnum

Supported spike sorter types.

ATTRIBUTE	DESCRIPTION
NULL	No sorter.
VEX_STREAM	Sorts a live stream in Allego.
SPIKES_STREAM	Not produced by current servers.
VEX_FILE	Sorts a recording's signals.
SPIKES_FILE	Sorts the spikes in a spikes fileset.