Answers are grounded in this paper and cite their sources.
Heme c, the phycobilins, biliverdin, the photoactive yellow protein chromophore and retinal are attached to their proteins by a covalent bond, which a PDB entry records as a covale row in _struct_conn. This note counts how often the bond is present in the coordinates and absent from the record, across every PDB entry containing one of these components (downloaded 3 November 2026). Not recorded: heme c 1,049 of 9,251 contacts (11.3%); phycobilins 3,885 of 13,972 (27.8%); biliverdin 19 of 263; retinal 31 of 916; PYP 1 of 90. For the thioether families the modelled bond length predicts most of it. In heme c, 999 of the 1,004 thioethers modelled at 2.0 Å or more have no record, against 1 of the 8,124 modelled between 1.6 and 2.0 Å. 234 heme-binding chains carry one heme, two thioethers in the coordinates and one record; in 226 of them the unrecorded bond is the one at 2.0 Å or more. Two further patterns are not explained by length: depositions that lack the record at normal bond length and come in sets, including a five-entry time-resolved phytochrome series in which the two illuminated structures record the anchor bond and the three dark structures do not; and a few cryo-EM phycobilisome models with thousands of cysteine–pigment contacts shorter than 1.6 Å. Many of the modelled lengths are a software default. In 118 entries the thioethers sit on 1.766 Å, and 52 retinal Schiff bases sit on 1.430 Å. These are the generic carbon–sulfur and carbon–nitrogen values that cctbx, the library under PHENIX, gives a linked pair at default settings, with a sigma of 0.01 Å. The restraints are reproduced on seven builds from 2020.8 to 2025.11, and none of the 755 entries that declare another refinement program is pinned on the thioether value. 988 unrecorded contacts sit on it. The supplement holds one table per family with one row per contact, the heme c parity table, provenance per entry, the scan script, the test fragments and restraint listings for every build, and the entry lists for the default values.
No takes yet. Share an insight, caveat, or question.
Michael T. Varman (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: