521 gaps · 136 records · 8 fields
What we do not know, enumerated
A blank field is a bug. A named absence is a finding. These 521 say which field is empty, on which record, and why — so the shape of what we hold can be checked rather than assumed.
Last updated
A gap is a claim, not a blank
Every row here asserts something: that we looked for a specific value on a specific record and did not find one that met the bar. That is different from a field nobody filled in. A blank invites you to supply the answer from the surrounding text; a named absence tells you not to. The grading rubric covers what qualifies as a source.
By field
Nothing we hold supplies it
- amino-acid sequence — 96 records
- CAS registry number — 53 records
- InChI — 46 records
- InChI key — 46 records
- molecular formula — 46 records
- molecular weight — 45 records
- PubChem identifier — 45 records
- SMILES string — 45 records
No source on file holds this
- amino-acid sequence — 15 records
- PubChem identifier — 10 records
- molecular formula — 9 records
- molecular weight — 9 records
The reference database returned nothing
Not applicable to the compound
By record
136 of the 185 records are missing at least one of the eight fields tracked here. The other 49 hold all eight.
This count is taken from the records themselves rather than from what a source admitted to. It used to be the other way round, and it read as a much better number: 166 records had no source-declared gaps and showed “not on file: 0” — including one holding none of the eight fields at all.
What this ledger does not cover
These 521 are the gaps on records we publish. An earlier version of this count said 1,515, and 1,227 of those belonged to 158 rows that were never compounds — scraped seller page titles ingested as records, including HTTP error pages, order-tracking screens and blog posts. They are excluded at the export step rather than filtered here, because data that cannot be published should not reach the page layer at all.
It covers eight fields, and only those: molecular weight and formula, CAS number, PubChem CID, amino-acid sequence, SMILES, InChI and InChI key. A record holding all eight is complete by this measure and not in any wider sense — no dosing, no pharmacokinetics, and nothing about evidence, which the rubric handles separately.
Until recently it also covered only what a source had declared missing, which made it a report on our sources' paperwork rather than on our data. 166 records had no declaration and showed “not on file: 0”, one of them while holding none of the eight fields. The rows are now counted from the records themselves, which is why this number went up rather than down.