FILE
HM–2026–06
STATUS
AUTHORITATIVE DESIGN
CLASS.
TECHNICAL · PUBLIC
MILESTONES
M0 → M7
STANDARD
MIQE · dMIQE
Agnostic
of the manufacturer: any real-time or digital thermocycler, normalised to a canonical model
8
milestones, from file capture to the issued report, auditable backwards
3
possible verdicts per sample and target: positive, negative or indeterminate, with the scope declared
0
thresholds written into the code: controls and limits live as configurable data
+
+
+
+
What makes it unique
Manufacturer-agnostic, and with the assay behind it. The thermocycler's own software interprets what its own machine measured; Hermes normalises anyone's file to a canonical model and interprets it knowing how that assay was designed: which primers, which target, what melting-curve prediction was made and which exclusions were declared.
That continuity is what does not exist today: the measured melting curve is compared with the one predicted at design time, the discrepancy becomes signal, and the system's knowledge improves with every real run instead of staying in the shift report.
01
The state of the art
Every machine, its own software. And the judgement, in the analyst's head.
A laboratory with three thermocyclers from three brands has three formats, three threshold criteria and three reports that do not compare with one another. Harmonisation is done by a person, by hand.
02
The chain of milestones
From the instrument file to the signable report.
Eight steps, each with its input and its output, so that a verdict can be audited backwards to the raw curve that produced it.
M0
File capture
Direct upload or automatic pickup from a watched folder in the laboratory, with an integrity check and manufacturer detection.
M1
Normalisation to a canonical model
Parsing the file according to its access tier — open interchange format, exported table or proprietary format — to produce a single representation of the run.
M2
Clean signal
Baseline correction and curve normalisation before measuring anything, because a threshold over an uncorrected signal measures noise.
M3
Quantification
In real time, threshold cycle by second derivative and efficiency by regression over the exponential phase. In digital, partition clustering and Poisson correction.
M4
Melting curve
Melting temperature from the fluorescence derivative and comparison against the curve the design predicted. The discrepancy is information, not noise.
M5
Run quality control
Validation of the controls in line with the reporting guidelines: no-template, internal amplification, positive and no-reverse-transcription, plus limits of detection and quantification.
M6
Verdict
Positive, negative or indeterminate per sample and target, with the scope declared. An honest indeterminate is worth more than a positive that does not hold up.
M7
Issuing
A report to the standard, export, notification over the event bus and filing in the laboratory's run library.
03
The boundary
What Hermes does and what belongs to others.
It is written so as not to overlap with anything: it interprets real thermocycler runs, full stop. It does not design, does not simulate and does not capture sequencers.
Yes
Read and interpret real runs from real-time and digital thermocyclers, from any manufacturer.
Normalise incompatible formats into a single model that allows comparison across instruments.
Validate the run controls and issue the verdict with its report to the standard.
Return to the system's knowledge the contrast between what was predicted and what was measured.
No
It does not design primers: that belongs to SnapPrime and the kit generator.
It does not do in silico PCR: biophysical prediction belongs to DigiCycler.
It does not capture sequencers: it reuses the existing ingestion infrastructure, adding its own platform adapter.
It does not invent thresholds: the control rules are curated, configurable data, not constants in the code.
04
Technical sheet
Input
An export from a real-time or digital thermocycler: open interchange format, exported table or proprietary format, depending on the manufacturer's access tier.
Output
A verdict per sample and target with its quantification, a report to the MIQE and dMIQE guidelines, structured export and an event on the system bus.
Ingestion
Direct upload or automatic pickup by an agent on a watched folder, with queueing and integrity, reusing the ecosystem's ingestion infrastructure.
Reuse
The design module's melting prediction as the reference curve, the control rules as curated knowledge, and the run library shared with the other modules.
Traceability
Every verdict can be walked backwards to the raw curve and to the design that produced those primers.
Status
Authoritative design approved in June 2026 after three passes of verified research on formats, licensing, mathematics and control rules. Implementation is under way as a chain of milestones.
Limit · access
The level of detail depends on what each manufacturer allows to be exported: with an open format the full run is recovered; with an exported table, whatever the table carries. The report declares which tier it worked with.
Limit · scope
Hermes interprets what the instrument measured. It does not fix a badly designed assay nor replace method validation: it makes visible, quickly and with a uniform criterion, what the run says.
Full name: Harmonized Evaluation of Real-time and digital-PCR Multi-instrument Export Signals. Real-time quantification by second derivative and efficiency by linear regression over the exponential phase. Digital PCR by density estimation and Gaussian mixture with Poisson correction. Melting curve by the negative derivative of fluorescence against temperature, checked against the design module's Poland-Scheraga prediction. Reports to the MIQE and dMIQE guidelines.
What the instrument measured, checked against what was predicted.
BIOTECNO.org · Vitoria-Gasteiz
EU + Codex