Emergent
Molecular Atlas
Explore the atlas
07 · Molecular evidence, practical decisions
Connect the molecule to the measurement.
Check what a gene can mean.
Pin down the sample and depth it came from.
Choose the test that settles it.
Three source-linked casesEvidence of 26 September 2026
Follow the evidence networkIllustrative mangrove workflow
Exact pairing still required
Available nowMolecular diligence and measurement planning
Next testDoes molecular evidence improve prediction?
08 · Explore the evidence
Every interpretation
has an evidence trail.
Behind these cases is a formal evidence model: an ontology whose rules a computer checks. It traces every statement to its source, keeps apart things that are easy to conflate, and lets a promising interpretation stay on hold with the reason in view.
What the evidence model will not allow
- A marker gene or RNA signal read as a methane flux
- A sequencing library treated as a physical sample
- An unassayed gene reported as absent
- A concentration relabeled as a flux
- A hypothesis promoted to a fact
- A carbon-credit approval asserted by the graph
For a project developer or verifier, the output is an evidence-readiness dossier, not a credit or a tonnage estimate. It shows which molecular evidence can answer which review question, and which measurement would resolve what it cannot.
Molecular Verification Ontology 0.2.0, built on the W3C RDF, OWL and SHACL standards. Each of its 18 fixed review queries was checked against the source tables. This public view shows three curated cases; the internal model holds selected evidence for 145 genome records and has not yet had independent scientific review.
Select a node to unfold its evidence.
Better interpretation. More purposeful measurement.
These reviews define which features and samples can enter a predictive-model test.
A molecular map now. A rigorous field test next.
144 sample-events = 36 plots × 2 microsites × 2 seasons. Revisits are repeated measurements, not independent replicates; the 3 plots at each stage and salinity position are the replicate unit.
Reference links are leads, and the map shows their limit.
Molecular payload counts retain the 10 August snapshot. The map and reference links were recomputed on 29 September 2026 after reconciling ESM-2 pooling to layer 33. The map draws a selection of links; ecological transfer and methane prediction still require validation.
The words behind the evidence
The report and the evidence explorer use a few technical terms. Each one names a specific evidence object or stage, so screening, comparison and future calibrated risk stay distinct.