Reproducible by construction
Genomarker
Upload your datasets, run publication-ready analyses, and collaborate across your institution — no bioinformatics team required.
The Platform
How Genomarker works: upload and ingest, exploration, differential expression, enrichment, and machine learning — with confounder checks enforced at submit.
Read more →Applications
Genomarker in practice: biomarker discovery studies, differential expression, calibrated classifiers, pre-registered analyses, and team collaboration.
Read more →Results
What a Genomarker result carries: a signed Reproducibility Receipt, a 0–100 Reproducibility Score, reporting checklists, and a DOI-minted evidence trail.
Read more →What is Genomarker?
Genomarker is a SaaS platform for reproducible biomarker discovery. Researchers upload expression data — bulk RNA-seq, microarray, and single-cell RNA-seq — and run publication-ready analyses, from PCA and clustering to differential expression and machine learning, without standing up a local bioinformatics stack. Every result exports with a signed Reproducibility Receipt attesting its provenance.
Why it matters
Biomarker discovery has a reproducibility problem: candidate markers that look decisive in one dataset too often fail in the next. The cause is rarely bad faith — it is invisible choices: an unmodeled batch effect, a leaky cross-validation split, an analysis plan quietly rewritten after seeing the data.
Genomarker’s answer is fewer ways to be wrong. The platform refuses to run a confounded design at submit time, screens every classifier for leakage and overfitting, and binds each verdict into a signed record. The wow is the stop, not the speed.
The catch
Most tools warn; Genomarker stops. Confounder Pre-Flight checks every submitted design for confounding and hard-refuses a confounded design — with the evidence and the fix. SAP-as-Executable turns the analysis plan into a pre-registered, hash-locked artifact with a Zenodo DOI, so every result is tagged pre-specified, post-hoc, or exploratory against the plan that came before the data. Rigor is the default, not the user’s discipline.
The platform never authors a result — you sign off on every choice.
Explore the platform
- The platform — the analysis pipeline: upload and ingest, exploration, differential expression, enrichment, and machine learning.
- Applications — what researchers use it for: biomarker studies, calibrated classifiers, pre-registered analyses, and team collaboration.
- Results — what a result carries: Reproducibility Receipts, Reproducibility Scores, reporting checklists, and DOI-minted evidence bundles.
Or watch a recorded demo of the platform in action.
Frequently asked questions
What is Genomarker?
Genomarker is a SaaS platform for reproducible biomarker discovery. Researchers upload expression data, run publication-ready analyses — from PCA and clustering to differential expression and machine learning — and export signed Reproducibility Receipts that attest the provenance of every result.
What can I do with it?
Upload expression data (up to 1 GB, GEO series matrices, microarrays, single-cell h5ad), explore it with PCA, clustering, and normalization, analyze differential expression with DESeq2 or limma-trend behind a Confounder Pre-Flight gate, run enrichment over Reactome, GO, and MSigDB Hallmark, train XGBoost and random-forest classifiers with calibration and decision-curve analysis — then export a signed Reproducibility Receipt with a DOI-minted evidence trail.
What is a Reproducibility Receipt?
A signed, hash-chained analysis record: pinned seeds, software versions, dataset fingerprint, and a canonical RFC 8785 hash, published at a public verify link. Receipts attest provenance — anyone can check exactly which data, code, and parameters produced a result.
What does it cost?
Genomarker is free during early access. Your data stays yours.
How do I get access?
Sign up at genomarker.vercel.app — public signup is open. The source repository on GitHub is currently private.
Try Genomarker
Genomarker is live and free during early access. Upload a dataset and run your first analysis in minutes — no bioinformatics team required. Your data stays yours.
The source repository is currently private.