Benchmarks you can inspect.
Small, traceable datasets and strict scoring tools for testing bioinformatics software. Start with a six-assay example, then bring results from your own pipeline.
Using OpenOmicsBench? Please star the repository. It helps others discover the project.
01 / What it does
A test collection with an audit trail.
OpenOmicsBench lets you validate bundled data, inspect its source and transformations, score your own outputs, and compare methods across runs. Export a checked RNA-seq input pack with aligned counts, sample metadata and a declared contrast, then check your own files before analysis. Native DESeq2, edgeR and limma result tables can be scored by naming the benchmark directly. Reports and evidence packages make a run easier to review or reproduce.
The 12 biological bulk RNA-seq objects come from seven Expression Atlas studies. The other fixtures are declared synthetic tests for sequences, variants, peptide identification, single-cell clustering, chromatin peaks and read-level mock taxonomy. They are software test cases, not clinical reference materials.
02 / Collection
Find a benchmark.
Search by assay, study, species or identifier.
Loading collection…
Each card links to the dataset directory, including its manifest and declared limitations. Check the individual manifest before using a benchmark in a study.
03 / Get started
From install to report in two commands.
The demo scores bundled reference answers against their own truth sets. It shows how the evaluator works; it is not a measurement of your method. For an RNA-seq test input, run omicsbench rnaseq pack my-rna-test --id rnaseq-005. Its count matrix, sample sheet and contrast are checked together; the example contains paired tumour and adjacent non-tumour samples.
$ python -m pip install openomicsbench
$ omicsbench demo my-first-runmy-first-run/report.html to inspect all six comparisons.04 / Use and cite