[Ninjani]

Summary

• Update README and task-based guides for querying, structures, similarities, representatives, search, and evaluation.
• Generate README counts from release tables and move release history to CHANGELOG.md.
• Add interface-site discovery for monomer structures and FASTA inputs, and simplify search modes.
• Add queryable score shards to reduce downloads for targeted similarity queries.
• Expose per-ligand and ligand-pocket validation independently of system aggregates; consistently exclude hydrogen isotopes from heavy-atom counts and occupancy/B-factor averages.
• Add chain-completeness metadata and derive missing values during entry updates while preserving strict validation.
• Make documentation CI reproducible with a small local release fixture.

[github-actions]

Coverage report

Click to see where and how coverage changed

File	Statements	Missing	Coverage	Coverage (new stmts)	Lines missing
src/plinder/core
release.py
src/plinder/core/index
query.py					307
utils.py					115
src/plinder/core/scores
custom.py					2917-2927, 3065, 3303-3323
interface.py
ligand.py
protein.py
query.py
search.py					121
store.py					81
src/plinder/core/utils
cpl.py
src/plinder/data/annotations
aggregate_annotations.py
get_ligand_validation.py					20
ligand_utils.py
ligand_validation.py
src/plinder/data/pipeline
collate.py
public_release.py					52
score_store.py					96, 181, 183, 226, 232, 271, 318-327, 338
update_entries.py
Project Total

This report was generated by python-coverage-comment-action

[tjduigna]

Thanks for the updates. Please address the existing-release update regression noted inline before merging. All required CI checks pass.

[tjduigna]

[P2] Backfill chain completeness before validating updates to existing releases

Please backfill chain_unresolved_fraction for retained legacy rows before this check. The documented update_entries flow keeps old_rows.*: releases produced by the base revision lack this column, so removal-only updates fail here. With incoming rows, UNION ALL BY NAME fills the retained rows with null, which also fails this comparison. The weekly-update fixture now supplies the field, so passing CI does not cover this case. Derive the legacy value from the existing count/length columns, preserve validation of supplied values, and add a base-format removal-only regression test.