Skip to content

v0.2.1

Latest
Latest

Choose a tag to compare

View all tags
@AlexandreAndr AlexandreAndr released this 24 May 23:15
· 10 commits to main since this release
v0.2.1
b49cd07

This release adds two new brainsets, a redesigned CLI configuration experience, and several utility improvements for pipeline authors. It also includes data-quality fixes to three existing pipelines and a round of legacy API removal.

New datasets

vollan_moser_alternating_2025 — Hippocampus and medial entorhinal cortex recordings from the paper Left–right-alternating theta sweeps in entorhinal–hippocampal maps of space. (#127)

neuroprobe_2025 — Full pipeline and Neuroprobe2025 dataset class for the Neuroprobe Benchmark, along with new documentation on how to use it. (#79, #134)

Breaking changes

Several legacy utilities have been removed. If your pipelines import any of the following, update before upgrading:

  • brainsets.utils.mat_utils removed (#119)
  • brainsets.utils.dir_utils removed (#120)
  • dandi_utils.extract_metadata_from_nwb removed — was undocumented and unused (#117)
  • brainsets.utils.split_*_epochs and generate_train_valid_test_splits removed from public API — too specific to the Allen VC Ophys pipeline and moved there (#126)
  • Deprecated data.set_*_domain() calls replaced with direct attribute assignment across all pipelines (#111)

Dataset classes

torch_brain.dataset.Dataset classes are now available for each brainset, making it easier to load processed data directly into PyTorch workflows without writing custom loaders. Classes added for: AllenVisualCodingOphys2016, ChurchlandShenoyNeural2012, FlintSlutzkyAccurate2012, KempSleepEDF2013, OdohertySabesNonhuman2017, PeiPandarinathNLB2021, PerichMillerPopulation2018. (#67)

CLI improvements

  • brainsets config show / config set — new subcommands to inspect and update your config. The dataset root now defaults to the configured processed_dir. (#122)
  • Shell completion — run brainsets --install-completion to enable tab completion for your shell. (#116)
  • brainsets prepare --list / -s / --single — previously internal flags, now public. (#137)
  • --download-only — flag is now visible in brainsets prepare --help. (#98)

Utility additions

  • calculate_sampling_rate added to utils (#138)
  • mne_utils expanded with robust channel extraction/remapping and a validated concatenate_recordings workflow (#107)
  • New bids_utils helpers to fetch/group recordings, validate availability, and load participants .tsv/.json sidecar files (#107)
  • New split utilities for inter-session, intra-session, and inter-subject train/valid/test splits (#83)
  • s3_utils enriched with additional helpers to fetch from public buckets (#78)
  • BrainsetPipeline runner now automatically creates raw_dir and processed_dir if they don't exist (#125)

Pipeline data fixes

Three pipelines have been corrected and their derived_version bumped to 2.0.0. Re-processing is recommended if you use these datasets.

Pipeline Change PR
churchland_shenoy_neural_2012 Corrected task metadata: experiment is maze reaching, not center-out reaching #131
odoherty_sabes_nonhuman_2017 Behavioral data now stored as RegularTimeSeries #138
pei_pandarinath_nlb_2021 Behavior signals regularized; NaNs inserted at missing timesteps #140

Bug fixes

  • Fixed broken dandi version in pipelines (bumped to 0.74.0) (#101)

New Contributors

All Contributors

@AlexandreAndr, @divyansha1115, @estefanysuarez, @glchau, @milosobral, @nandahkrishna, @oumgharbi, @ShivashriganeshMahato, @TomGeorge1234, @vinamarora8

Full Changelog: v0.2.0...v0.2.1

Contributors

ShivashriganeshMahato, vinamarora8, and 8 other contributors
Assets 2
Loading