VollanMoserAlternating2025#

class torch_brain.datasets.VollanMoserAlternating2025(root=None, recording_ids=None, transform=None, dirname='vollan_moser_alternating_2025', **kwargs)[source]#

Bases: torch_brain.datasets.mixins.SpikingDatasetMixin, torch_brain.datasets.dataset.Dataset

Neuropixels recordings from MEC and hippocampus in rats during spatial navigation and sleep.

Rats performed various navigation tasks (open field, linear track, M-maze, wagon wheel) while neural activity was recorded from medial entorhinal cortex (MEC) and/or hippocampus (HC) using Neuropixels probes. Sleep sessions with identified SWS and REM epochs are also included. The dataset contains grid cells, head direction cells, and other spatially tuned neurons.

Preprocessing

To download and prepare this dataset, run

brainsets prepare vollan_moser_alternating_2025

Tasks: Open Field, Linear Track, M-Maze, Wagon Wheel, Sleep

Brain Regions: MEC, Hippocampus

Dataset Statistics

  • Subjects: 19

  • Total Sessions: 51 (31 Open Field, 7 Linear Track, 2 Wagon Wheel, 1 M-Maze, 1 Novel Open Field, 9 Sleep)

  • Recording Tech: Neuropixels

Navigation sessions (42 sessions) contain:

rec.spikes

Spike timestamps and unit indices.

rec.units

Per-unit metadata: id, location (mec/hc), probe_id, shank_id, shank_pos, mean_rate, ks2_label, is_grid.

rec.samples

All variables on the shared 10 ms timebase (speed-filtered at 5 cm/s) in a single flat IrregularTimeSeries:

Observed tracking variables:

  • x – head x-position relative to arena centre (m)

  • y – head y-position relative to arena centre (m)

  • z – head z-position relative to floor (m)

  • hd – 2D head direction / azimuth (rad)

  • speed – horizontal head speed (m/s)

  • theta – instantaneous theta phase (rad)

  • id – decoded internal direction (rad, from the LMT model)

LMT decoded variables (for each population {pop} in mec, hc, mec_hc):

  • lmt_{pop}_theta – theta phase (fixed; largely duplicates theta)

  • lmt_{pop}_hd – head direction (fixed; largely duplicates hd)

  • lmt_{pop}_id – internal direction (latent)

  • lmt_{pop}_pos_x – decoded x-position (latent)

  • lmt_{pop}_pos_y – decoded y-position (latent)

Populations not present for a given animal are NaN-padded so all sessions share the same field schema.

rec.theta_chunks

Theta-cycle-binned timeseries (separate timebase, one sample per theta cycle): id, L (log-likelihood, n_cycles x 30), P (probability, n_cycles x 30).

rec.probe_channels

Probe channel geometry: probe_id, channel_index, x_um, y_um, shank_id, connected.

Sleep sessions (9 sessions) have a fundamentally different structure and should be treated independently. They contain only:

rec.spikes

Spike timestamps and unit indices within SWS/REM epochs.

rec.units

Minimal unit metadata (id only).

rec.domain

Union of SWS and REM epoch intervals.

None of the navigation fields (samples, theta_chunks, probe_channels) are present on sleep sessions.

References

Vollan, A. Z., Gardner, R. J., Moser, M.-B. & Moser, E. I. Left-right-alternating theta sweeps in the entorhinal-hippocampal spatial map. Dataset: EBRAINS.

Parameters:

  • root (Optional[str]) – Root directory for the dataset. Defaults to processed_dir from brainsets config.

  • recording_ids (Optional[list[str]]) –

    Recording IDs to load. Defaults to all sessions. Can be:

    • A list of individual recording IDs.

    • A string shorthand: "all" (default), "sleep", "navigation", "of", "lt", "mmaze", "ww", "of_novel".

    • A RECORDING_IDS sub-group for finer control:

      from torch_brain.datasets.VollanMoserAlternating2025 import RECORDING_IDS
ds = VollanMoserAlternating2025(root, recording_ids=RECORDING_IDS.navigation.of)

  • transform (Optional[Callable]) – Data transformation to apply.

  • dirname (str) – Subdirectory for the dataset. Defaults to “vollan_moser_alternating_2025”.

get_sampling_intervals(split=None)[source]#

Returns a dictionary of sampling intervals for each recording. This represents the intervals that can be sampled from each session.

This dictionary will be used by torch_brain’s Samplers to know where to sample from.

The default method returns intervals containing the entire domain of each recording. This behavior can be overridden by subclasses to give out custom sampling intervals.

Returns:

Dictionary mapping recording IDs to their time domain intervals.

apply_namespace(data, namespace)#

Apply a namespace prefix to specified nested attributes in the data.

This method modifies the data object in-place by prepending the namespace to string attributes or string arrays specified in namespace_attributes.

Can be overridden by subclasses to apply the namespace in a custom way.

Parameters:

  • data (Data) – The Data object to modify.

  • namespace (str) – The namespace prefix to prepend (e.g., “experiment1/”).

Return type:

Data

Returns:

The modified torch_brain.data.Data object (same instance, modified in-place).

compute_average_firing_rates()#

Compute and return the average firing rates for all units in the dataset.

Returns:

DataFrame indexed by unit ID, containing a column ‘firing_rate’ (Hz)

with the average firing rate for each unit in the dataset.

Return type:

pd.DataFrame

get_recording(recording_id, _namespace='')#

Get lazy-loaded torch_brain.data.Data object for a recording.

Parameters:

  • recording_id (str) – The ID of the recording to load (same as from recording_ids()).

  • _namespace (str) – Optional namespace prefix to apply to attributes.

Return type:

Data

Returns:

Lazy torch_brain.data.Data object containing the full recording.

get_recording_hook(data)#

Hook method called after loading a recording in get_recording().

Subclasses can override this method to perform custom processing on recordings after they are loaded but before they are returned.

Parameters:

data (Data) – The Data object that was just loaded.

get_unit_ids()#

Return a sorted list of all unit IDs across all recordings in the dataset.

Return type:

list[str]

property recording_ids: list[str]#

Sorted list of recording IDs in the dataset.