Applications

Genomarker is built for researchers and institutions who need publication-ready genomics analysis without a dedicated bioinformatics team. Its main use cases:

End-to-end biomarker discovery

From raw expression matrix to a signed evidence bundle in one workspace: upload, QC and exploration, differential expression, enrichment, and a predictive model — each step recorded, each result exportable with its provenance attached.

Differential-expression studies

Bulk RNA-seq, microarray, and single-cell RNA-seq studies run through assay-matched estimators — DESeq2 for counts, limma-trend for arrays, pseudobulk or MAST for single-cell — with covariate and repeated-measures designs, and a confounder check before anything runs.

Classifier development

Teams building diagnostic or prognostic signatures train XGBoost and random-forest models with isotonic and Platt calibration, Brier score, decision-curve analysis, and subgroup AUROC — the evidence reviewers ask for — plus holdout and external-validation detection to catch overfitting before a journal does.

Pre-registered studies

SAP-as-Executable turns the analysis plan into a locked, hashed artifact with a Zenodo DOI before the first run. Every downstream result is tagged pre-specified, post-hoc, or exploratory against that plan — for grant applications, regulatory submissions, and collaborations that need to show the plan came first.

Team collaboration

Multi-tenant workspaces with org roles, shared projects, and scoped API keys let labs, cores, and institutions work on the same datasets and analyses with auditable access.

Target and mechanism review

The Gene Evidence Card assembles what is known about a gene from Open Targets, UniProt, STRING, ChEMBL, DGIdb, and AlphaFold — connecting candidate markers to druggability and mechanism evidence without leaving the platform.

Methods comparison

Methods-Diff compares an analysis — receipt against receipt, or receipt against a published paper via Europe PMC, Crossref, and arXiv — to show exactly how two methods sections differ, for reviewers, replications, and studies that must match prior work.