Trust center

Open evidence with an explicit scope

Review what is available, where the results come from and which limits accompany each deliverable.

Demo · public or synthetic dataThis site presents synthetic demonstrations and exploratory reanalyses of public data. Each result identifies its origin and method. The workspace supports research review, with clinical decisions remaining with the responsible team. Private-data processing requires a separate enablement and review.

What Nexalel is

An evidence layer for research in genomics and microbiome science.
A hosted, invitation-only workspace where verified pilots already run on public cases and synthetic tables.
A versioned method contract, with published artifacts, hashes and seeds.
Public cases computed on open data: a published soil reanalysis under CC BY and a genomics evaluation on GIAB and 1000 Genomes.
A report that separates observation, inference and interpretation.

What Nexalel is not

Not offered as a medical device or for clinical use.
Does not issue diagnoses or prognoses.
Does not classify variants for management decisions.
Does not recommend treatments or diets.
Does not deliver validated functional prediction.
Does not deliver trained machine learning models.
Does not combine host and microbiome into a health score.
Has no public customers or completed pilots we can cite.

Capability matrix

Each capability has one of three states. For those not yet available, we state what it would take to enable them.

Capability	State	What exists today	What it would take to enable it
Public cases	Available	A soil reanalysis on open data (Figshare, CC BY), a genomics evaluation on two public cases (GIAB and 1000 Genomes) and a synthetic microbiome challenge that verifies the software.	Nothing to examine them today. The human microbiome reanalysis on open cohorts will be published once its reuse rights are settled.
Versioned method contract	Available	Frozen versions, an identified container, fixed seeds and one hash per artifact.	For a new dataset, a review of compatibility, rights and controls before the method is agreed.
Genomics evaluation	Available	A hosted evaluation pilot on two public cases, with prepared annotations and fixed versions. No variant corresponds to a patient.	For your own data, usage agreements with the annotation sources and a review of the dataset, signed by contract for each project.
Hosted workspace	Available	API, queue, storage and worker deployed and verified end to end in pilots with partner accounts. Access is by invitation.	Private-data upload and self-service remain out of scope until agreed in writing.
Clinical interpretation	Not available	Not offered.	Out of scope. It would change the intended use and, with it, the regulatory framework.
Validated functional prediction	Not available	Not offered.	Independent experimental validation, which we do not have today and do not claim to have.
Integrated host and microbiome score	Not available	Not offered.	Published evidence that supports the integration. Today this is an open research question rather than a product.

Method contracts

These are the exact versions used to compute the public case. They are frozen and declared one by one.

The cohorts share the downstream analysis on counts, but not the upstream processing. The sensitivity review records its own environment and versions separately.

Nexalel synthetic challenge

nexalel-review-challenge.1729

Data generated to verify the software, its sensitivity and its design warnings. It does not represent patients and does not constitute clinical validation.

Method version: Snakemake · microbiome-counts-analysis 1.0.0 · OLS on CLR · BH
Seed: 1729
Artifact hash: 0defd2aa6fd09c7db252ad975103212debbab707257753d89e12ed02166e5625

Download the synthetic tables and test the analysis ↓ · 0.01 MB · No account, no form and no sales call.

What held up

A diversity signal holds under the documented depth check.

What did not hold up

Of the two significant contrasts, neither survived the sensitivity test. We publish that too.

We publish both. A method that only shows its successes cannot be evaluated.

Reviewable data cards

These cards describe the synthetic demonstrations only. The public microbiome case has its own report, manifests and limits. No hash constitutes biological validation.

Genomics · results view

A product fixture written for the interface. It does not come from a fresh pipeline run.

Version: preview-results.v1
Source: Nexalel · synthetic demonstration data
Reconstruction available: Rendering and export of the fixture, with no biological reconstruction.
Fixture SHA-256: 5e2c4e9e67d6362909f20eb242f58aa6f2a7ff85a6101abeafcb4930027ebb1b

What is missing

Reconstruction of this view from the original VCF, not yet demonstrated
Real biological annotations
Clinical validation

Explore this demo

Microbiome · illustrative composition

Three fictional time points and six illustrative groups. It corresponds to neither a cohort nor a sequencing run.

Version: nexalel-microbiome-demo-1.0.0
Source: Nexalel · synthetic demonstration data
Reconstruction available: Literal composition; Shannon = -sum(p * ln(p)); network not estimated because the cohort is too small.
Fixture SHA-256: a58f6d9b8a1a6b49b72b7a06af5f6cee51a956e70fb50a256055d9d303b6c24c

What is missing

FASTQ, 16S region or shotgun protocol
Negative controls and sequencing depth
Taxonomic reference and its version
Laboratory and batch metadata

Explore this demo

Download the data cards and reconstruction manifest as JSON ↓

Data handling

What we accept

On the site, public or synthetic data only. Your own data is accepted only after a review of the rights, purpose and controls of the specific dataset.

What we reject

Clinical, patient or identifiable genetic data, and files of uncertain origin. No form accepts files, and they should not be sent by email either.

Retention and deletion

Location, access, retention, backup and verifiable deletion are agreed in writing before anything is processed.

What never appears in the logs

File contents, variant rows, passwords, tokens, presigned links and client-visible object keys.

Explicit originPublic or synthetic data
Visible methodsSources and limits per result
Scientific reviewPublic case with sensitivity analysis

Controls and open items · Privacy · Terms

Regulatory framework

In Colombia, whether software counts as a medical device depends on its use, application and intended purpose. The label "research" on its own does not establish a regulatory exemption.

Nexalel declares a research use and excludes clinical decisions about people. We do not claim registration, certification or an exemption determination by the authority. Any change of use requires a specific legal and regulatory assessment.

INVIMA · intended purpose of software ↗

Status of the hosted workspace

Available

It is deployed and verified through pilots on public cases and synthetic tables, with invitation-only access and no self-service. Private-data upload is not enabled, so each study is contracted as a service with an agreed scope.

Methods and critical reading

Relative abundance is compositional, so a percentage does not measure absolute load. Diversity is not the same as health, association is not the same as causation, and a functional prediction is not measured activity. The genomics and microbiome demonstrations are never combined into a diagnosis.

These references support the methodological limits and neither validate nor commercially endorse Nexalel. Content reviewed: 2026-09-09.

What does your next study have to demonstrate?

We first agree on the question, the assay, the data rights and the deliverable. Then we decide whether working together makes sense.

Discuss a study See the public case

Back to home