Backfill: opening the index
The index opens with a retrospective quarter. What was accepted, what was refused, and the four problems that were obvious from the first hundred records.
Open AssayIndex opened for submissions in 2024-Q1 with a backfill: reports assayed before the index existed were accepted retrospectively where the submitter could document custody and name the laboratory. That is a weaker evidentiary standard than the index applies now and every 2024-Q1 record carries a note saying so.
What was accepted and what was not
Three criteria were applied to backfilled submissions. A named laboratory. A lot number. A documented interval between purchase and assay. Submissions failing any of the three were refused, and the largest single category of refusal was the screenshot: a photographed or cropped chromatogram with no lot number and no laboratory letterhead. Those are not records and the index does not hold them.
The four problems visible from the first hundred records
- Purity is reported and nothing else is. Of the opening cohort, the overwhelming majority carried a purity figure alone. No water, no counter-ion, no content, no endotoxin. A purity figure is the cheapest measurement available and it is the one that says least about what is in the vial.
- Methods are not stated. Gradient duration, column, wavelength and reporting threshold are absent from most certificates. Without them a purity figure cannot be compared with another purity figure, which makes an index of purity figures a harder object to build than it first appears.
- Certificates do not close arithmetically. The first hundred records surfaced several certificates whose declared purity and declared total impurities did not sum to 100, and one in which the largest single impurity exceeded the total. Those observations became rules AX-R02 and AX-R03.
- Provenance is undocumented. Whether a sample was bought or supplied was not recorded on most submissions and had to be reconstructed from correspondence. It is now a required field and it is the single most useful thing the index added in its first year.
| Compound | Records | Median purity | Suppliers |
|---|---|---|---|
| Tirzepatide | 30 | 98.70 | 4 |
| Semaglutide | 24 | 98.86 | 3 |
| Afamelanotide | 17 | 98.24 | 2 |
| PT-141 (Bremelanotide) | 16 | 99.60 | 2 |
| Semax | 15 | 99.36 | 2 |
| N-Acetyl Semax Amide | 14 | 99.25 | 1 |
| Retatrutide | 10 | 98.95 | 2 |
| IGF-1 LR3 | 10 | 98.95 | 1 |
| Desmopressin | 8 | 99.77 | 1 |
| TB-500 | 7 | 99.46 | 2 |
| GHRP-6 | 5 | 99.17 | 1 |
| Melanotan II | 3 | 99.70 | 1 |
Method decisions taken this quarter
- The index reference method RG-2 was defined, so retention windows could be stated against a common scale.
- Monoisotopic mass was adopted as the theoretical basis for every ppm calculation, after several backfilled records were found to be comparing observed monoisotopic values against theoretical average masses — a 600 ppm error that looks like an identity failure.
- Type 7 quantiles were adopted for every median and interquartile range on the site, and the choice is documented rather than left implicit.
- A reporting threshold of 0.05 % area was adopted, matching ICH Q3A at moderate dose.