Research AI with a record behind every result.
Operon gives a research team one private place to ask questions of curated biological knowledge, run analyses and validated pipelines, and bring in partner methods. Every answer cites the evidence it rests on. Every result carries the data, method and version that produced it.
Someone has an idea. Nobody knows how it got there. Later, you're trying to reconstruct where it went wrong. Operon is the antithesis of that.
40+ tools, each one recorded when it runs.
The research agent doesn't answer from memory. It calls tools, and every call becomes part of the record. Here is what those tools do.
- —Which diseases a microbe is linked to, and the evidence
- —What a microbe produces, and which drugs act on it
- —A disease's microbiome signature
- —The neighborhood around any entity
- —Mechanistic paths from a microbe to a disease
- —Metabolite production chains
- —Bridge taxa shared between diseases
- —Under-explored disease pairs worth a look
- —PubMed, bioRxiv, ClinicalTrials.gov and Semantic Scholar
- —Gathering evidence for and against a claim
- —Checking for known confounders
- —Analyses and custom pipelines
- —Nextflow pipelines, nf-core included
- —Campaigns: batches of experiments run as one
- —Partner methods, through a registry
- —A review pass before a result is trusted
- —Adding findings to your Company Graph
- —Retracting a finding, recorded too
- —Recording runs made outside Operon
A catalog of analysis steps, and the Nextflow ecosystem.
Steps chain into pipelines. Anything costly is planned first and launched only once someone approves it.
- Alpha diversity
- OTU clustering
- Taxonomic classification
- Sequence alignment
- Network inference
- Correlation networks
- gLV calibration and simulation
- Ecosystem modelling
- Guild ecology
- Perturbation prediction
- Intervention profiles
- Pathway enrichment
- GraphMiner rule mining
- Statistical summaries
- Files and database queries
- Column selection and transforms
- Plots
- Write-back to your Company Graph
Plus any Nextflow pipeline on your approved list, nf-core included. Operon launches it, tracks it, and records the finished run, failed tasks and all. Partner methods run through a registry and are recorded the same way.
A real record, from a real run.
This is Operon recording an actual run of the public nf-core/demo pipeline: eight tasks, seven completed and one failed. The record keeps the failure. Pipeline, revision, owner and compute time are read from the run's own output, not typed in by hand.
Every record carries a trust label: reproducible when Operon ran it, attested when it was read from a real run like this one, declared when a partner reported it. Inputs and outputs are content-addressed, and each record links to the steps it came from, so any result traces back to the question that started it.
- Trust label
- Attested: read from the run's own output
- Pipeline
- nf-core/demo, revision 32893afef8
- Run by
- nf-core (the pipeline's owner)
- Engine
- Nextflow 25.10.7
- Tasks
- 8 · 7 completed · 1 failed (exit 2)
- Compute
- 98 seconds
One organization, one deployment.
Your data and results stay in your deployment. MicroMap supplies the curated reference your Company Graph links to.
Runs where your data already lives.
Deployed for your organization alone: hosted by Graphomics or in your own cloud, on Kubernetes or Docker Compose.
Bring your questions. See them answered in Operon.