The life of a test · 06 Records and proof
Every result carries its proof.
Who did what, when and why; which engine, version and settings made each result; and the evidence that the method works.
Validation
Accuracy, measured and published.
Public truth and the laboratory's own comparison results stay separate, with the method, scope and every disagreement open to review.
- Validation packs for each released method version
- Synthetic evidence labelled as software evidence
- Laboratory concordance required for automatic qPCR calling
- Versioned workflows, so the method is the same every time
Records
Every action, on the record.
Who, when, the value before and after, and why. Each entry holds the hash of the one before, so an altered entry is found.
Plate record · ITGA2B cats
24 entries, none altered
- 01
Imported
sample sheet cat-sheet.csv, 22 wells, 23 files kept unchanged · Analyst 8, 12:43:23 UTC
86d9239eb4after nothing
- 02
Kept no call, back for sequencing · G12
chrE1:44416063 C>CGG no call · Analyst 8, 12:45:26 UTC
4b767bf484after 86d9239eb4
- 04
Confirmed in the sweep · A5
chrE1:44416063 C>CGG homozygous reference · Analyst 8, 12:45:26 UTC
fab6285eebafter 9e67a55f94
- 24
Released
version 1, single analysis: 20 results locked, 2 back for sequencing, no controls · Analyst 8, 12:45:26 UTC
962b848a41after 2384ca08bd
- Every plate, run and test version recorded
- The whole record checked in one step
- Kept in the lab's full export, and restored with it
Your data
It leaves exactly as it was kept.
Open formats with a manifest, every file read back and compared after it is written, and a lab that restores as itself.
- Every original byte for byte
- FASTA, GenBank, AB1, RDML and CSV exports
- The whole lab exported and restored