HyDB — an observation corpus and sequence registry for PET hydrolases

HyDB is a provenance-tracked corpus of experimental observations on poly(ethylene terephthalate) (PET) hydrolases with the sequence registry they are anchored to. python3 VERIFY.py - standard library only - 22 property checks, 0 failures. Contents (1.6, internal build 0.1.117): 40,929 observations across five views - PET 28,094, MHET 1,317, SYS 5,056, CORE 6,289, AVAL 173; 12,147 at the direct-PET counter; 6,040 sequence records carrying 5,946 distinct normalised-residue fingerprints; 313 sources; 46 acquisition rounds. What changed in 1.6. Two reproducibility defects are closed. REPRODUCE.md and requirements.txt are now GENERATED by the build. They were written by hand into versions 1.4 and 1.5 and the next rebuild erased them without a trace, so the package shipped twice announcing instructions it did not contain: a file written by hand inside a generated tree does not survive, and its absence is invisible. And code/semantic_attribution.py no longer carries an absolute path to the author's machine; asked to run without HYDB_SOURCES, it now states precisely which inputs it needs and why they are not redistributed, instead of searching the reader's disk. Carried over from 1.4 and 1.5: frozen PlasticDB and PAZy snapshots in external/ with URL, retrieval time and SHA-256, from which code/external_comparison.py recomputes every overlap figure; the complete validation traces in validation/, including the 120-row transcription sample with its per-row verdict and retrieval profile, the 238-row semantic-attribution sample, and the UniProtKB reconciliation with its query inputs and frozen responses; one shared rule for reading conditions in code/conditions.py, with a test that fails if a plotted number stops coming from it. Reading the data. An empty condition cell means not reported, never zero. counts_direct_pet is not a boolean - test it for the exact string YES; the negative case is written NO, FALSE and NO_BY_VIEW for historical reasons. Version numbering. Zenodo version 1.6, following 1.5 (10.5281/zenodo.22100243), which remains available and unaltered. The archive filename carries the internal curation build number, 0.1.117. Archive: HyDB_public_v0.1.117.zip, 26650423 bytes, sha256 9c9c8255bd0728dfe301019e24ef1be033b24ed0abe27bc795030abe56d092da.