Changelog

1.0.0

Changed

• Breaking: get_skeleton now returns an ossify.Cell instead of a meshparty meshwork. The skeleton is at nrn.skeleton, the level 2 graph at nrn.mesh, and synapse annotations at nrn.annotations.pre_syn and nrn.annotations.post_syn.
• Breaking: Renamed the get_skeleton synapse_reference_tables argument (a dict) to reference_tables (a list of table names), matching the ossify API.
• Breaking: Replaced the pcg_skel/meshparty dependency with ossify for skeleton retrieval.
• Breaking: Dropped Python 3.10 support; the minimum supported version is now Python 3.11.
• Upgraded to standard-transform 2.0.0 and moved off its deprecated transform_nm/streamline_nm accessors. Streamline straightening is now anchored at each neuron's root/soma.
• Bumped minimum dependency versions (caveclient>=8.2.1, cloud-volume>=12.14.1, nglui>=4.7.4) and promoted cloud-files and tqdm to direct dependencies.

Added

• Added a vertex_lookup optional extra for the mesh-to-skeleton vertex lookup tooling (VertexAssigner), installable with pip install "cortical-tools[vertex_lookup]". This bundles gpytoolbox (which can be slow to build), joblib, and tqdm-joblib, which are no longer installed by default. Using VertexAssigner without the extra raises an informative ImportError.

0.1.1

Fixed

• Fixed docs GitHub Action to pin mkdocs<2.0 and update mkdocstrings-python 2.x configuration (import → inventories, removed deprecated allow_section_blank_line option).
• Fixed Python 3.10 compatibility: Self from typing is only available in 3.11+, now imported from typing_extensions on 3.10.

0.1.0

Added

• Lifted a series of commonly-used caveclient methods directly onto the dataset client, making them accessible without going through client.cave. This includes methods for chunkedgraph operations (get_roots, is_latest_roots, get_latest_roots, suggest_latest_roots), skeleton retrieval (get_skeleton, skeletons_exist), L2 cache (get_l2_ids, get_l2data), info service (get_datastack_info, image_source, segmentation_source, viewer_resolution, image_cloudvolume, segmentation_cloudvolume), JSON state (get_state_json, upload_state_json), annotation staging (stage_annotations, upload_staged_annotations), and materialization utilities (most_recent_materialization_version, version_timestamp, latest_valid_timestamp).

0.0.7

Added

• Added root_ids argument to neuroglancer_url method to allow specifying root IDs to include in the Neuroglancer view.

Fixed

• Fixed repr methods for MicronsProdClient to return strings correctly.
• Updated nglui dependency version to fix string issues.

0.0.6

Fixed

• Fixed a bug in root_id_to_cell_id that could cause incorrect results in certain conditions when using alternative lookup tables.
• Removed a non-available alternative lookup table from the v1dd_public dataset client.

0.0.5

Added

• Added cortical_tools.load_client method that takes a dataset name (currently one of "v1dd", "v1dd_public", "microns_prod", "microns_public") and returns the corresponding dataset client. This is intended for scripts and paramterized notebooks.

0.0.4

Added

• Added query_synapses method to query synapses inclusively with reference tables.
• Added optional bounds argument to get_l2_ids to limit search area.
• Added dataset-active tests for all datastacks.

Fixed

• Fixed bug in streamline transformations for skeletons and synapses.
• Fixed bug in get_l2_ids that did not work.
• Allowed cell_id_to_root_id and root_id_to_cell_id to work with a single numeric ID.
• Suppress caveclient warnings.

0.0.3

Fixed

• Improved mesh vertex lookup memory usage and performance. Should no longer crash on large meshes due to out of memory issues.

0.0.2

Changed

Added additional docstrings.

0.0.1

First release!