PETHyDB: a condition-resolved, provenance-tracked PET-hydrolase activity resource

A harmonised integration layer over the published literature on poly(ethylene terephthalate) (PET) hydrolase activity. Contains 1,594 validated enzyme sequences and 32,820 activity observations re-expressed in a controlled evidence vocabulary of six orthogonal axes (target, biological level, detection method, value type, role, state), with provenance to the primary source table and row. Key design principle: direct-PET status is derived from the axes rather than asserted, yielding 9,009 canonical direct-PET observations under a mechanical, user-reproducible rule. 6,047 measured zeros are preserved as a distinct state. Whole-cell, dual-enzyme and process-level measurements are never attributed to a single purified sequence. Derived sums are never counted alongside the species they summarise. Model substrates (pNP esters, BHET-only, fluorescence proxies) are excluded from direct-PET by construction. Scope: this resource is an integration layer, not a republication of third-party raw measurements. Where source licences permit (CC-BY, open deposits) values are included; otherwise the harmonised derived record is provided with a pointer to the original article. Draft version 0.1.0. Version note (0.1.79, 3 August 2026): 33,118 observations, 12,083 canonical direct-PET, 2,453 sequences, 156 sources, 54 distinct quantity kinds. A composite package offered three sources, two already held: one sub-package was the round-11 delivery byte for byte, and the Seo Science source already had 289 observations and 341 sequences here. Two of its numeric claims were misattributions: a melting temperature of 68.6 degC assigned to Mipa-P belongs to node C158, and 69.5 degC assigned to Kubu-P belongs to the benchmark Thc_Cut2. Neither number appears anywhere in the 151-page supplement, which prints the real values — 68.3 degC for Mipa-P and 87.7 degC for Kubu-P. Assigning 69.5 to an enzyme the source prints at 87.7 is an 18-degree error and a measurement filed against the wrong protein. What the package summarised in two qualitative lines is here in full: the complete engineering ladders of Figures S17 and S18, ten Tm steps for Mipa-P from 68.3 to 92.4 degC and eleven for Kubu-P, six of them printed as > 99.9 degC and kept as lower bounds rather than exact values. Three conformational abundances from ACS Catalysis are recorded as simulation output, as the article itself states. The round adds no rows to the direct-PET counter, which is the correct outcome. All four tables reproduce byte for byte from the deposited archive alone. Deposit remains a draft.