Repository navigation
Serve the 2026-09 release from Cameo and execute all docs notebooks against it - #126
Merged
Merged
Conversation
tjduigna
changed the title
tjduigna
force-pushed
the
release-date-prefix
branch
from
September 30, 2026 16:01
50538ae to
b975dd3
Compare
- restore the original physical schema when DuckDB rewrites a patched alignment shard (DuckDB writes string; pandas 3 writes large_string) - assert missing string values with isna() so the query test holds for pandas 2 (None) and pandas 3 (NaN)
…nts_data.txt - dev dependency group is the complete CI/dev environment (adds plots extra and semver); networkit must come from a wheel via no-build-package - commit uv.lock and install with uv sync --locked into the Conda env that provides the conda-only tools (openstructure, mmseqs2, foldseek) - build release distributions with uv build - pin the project Python to 3.12 with .python-version
…tion Published CIGAR shards omit alignment scores, so Scorer.load_alignments reads the matching exports/protein_similarity_scores shard. prefetch only downloaded the CIGAR shards, so reconstruct_similarity_scores failed on a lazily populated cache. Register the score shard as a release path and fetch it when include_protein_scores=True (used by ligand reconstruction only).
- drop the PLINDER_RELEASE=2026-07 / PLINDER_RELEASE_NUMBER=1 overrides so notebooks use the default release and download lazily from R2 - remove the notebook execution skip list - install docs builds with uv sync --locked --extra docs into the Conda env (OpenStructure and MMseqs2 are needed by the evaluation and search notebooks)
Coverage report
Click to see where and how coverage changed
This report was generated by python-coverage-comment-action |
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
…numbers Custom-scoring alignments keep their CIGAR column but are already mapped to query/target residue numbers. load_alignments only checked for decoded positions, so it tried to decode them, which needs a release chain lookup the custom scorer doesn't have (ValueError: CIGAR alignments require a chain lookup). Found by executing docs/examples/custom_scoring.ipynb.
tjduigna
changed the title
CI now tests the dependency set users get from pip, catching upstream breakage early rather than through a lockfile that must be refreshed.
…lease-date-prefix # Conflicts: # src/plinder/core/release.py
- make_sub_dbs now also builds monomer databases whenever holo is built; when no monomer chains exist (e.g. the end-to-end test entry) linclust segfaults on the empty database. Skip monomer databases with no chains; downstream bundling already treats them as optional. - check query_batch_index is not None so mypy can narrow the batch slice.
The end-to-end entry has monomer chains, but none are selected into the MMseqs sub-database, so the chain-count check in make_sub_dbs never fired and linclust still crashed. Check the selected database instead and skip clustering an empty monomer universe; revert the earlier chain-count check.
- README, getting-started and migration pages now point at 2026-09 on plinderdata.org instead of 2024-06/v2 and the release-number layout - regenerate the release tree in docs/dataset.md from the published manifest (adds protein_clusters, ligand_pocket_residues, apo/interface_apo CIGARs and the manifest; drops ligand_archives/manifest.json and per-fragment ligand_scores) - document the manifest builder's prefix argument and the publish order / Cache-Control needed for manifest updates
- directory downloads drop cached Parquet files the release no longer lists, so dataset readers never mix removed fragments (e.g. the 290 old ligand_scores files removed from 2026-09) with their replacements - skipping an empty monomer sub-database also retires an earlier build, so a stale monomer target is never published as current - custom scoring treats monomer search databases as optional per backend, and public_release no longer requires monomer artifacts, matching the pipeline - monomer chain masks cast fillna results to bool for pandas 3 - build_r2_manifest skips a root folder placeholder; clarify its prefix argument
tjduigna
changed the title
Ninjani
approved these changes
Oct 9, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
The current
2026-09dataset is hotfixed on the Cameo file server. This PR makes the package use that release directly by default, so users can install this branch and get the current data without a hosting override.Release access
DEFAULT_RELEASE = "2026-09"; remote releases usehttps://cameo3d.org/plinder/PLINDER-<YYYY-MM>/. The local cache stays at<mount>/<bucket>/<YYYY-MM>/.PLINDER_MIRROR_URLcan select another file-server root with the same layout and directory indexes.Cameo does not publish per-file checksums. Transfers validate HTTP Content-Length rather than MD5. Publishers must update Last-Modified when replacing a file; a same-size change preserving that timestamp requires
data.force_update=true.Other changes retained in this branch
fix/major-refactor-bugs, including monomer search support and public-release staging.Validation for the Cameo change
index/entry_sources.parquet(259,015 rows), downloaded/read bothprotein_clustersfiles (1,093,677 rows each), traversed release and nested export directories, and confirmed that unchanged cached files are not transferred again.