A professional-grade genomics workbench (genome browser, plasmid designer, CRISPR studio, ancestry & pharmacogenomics, and polygenic-risk modeling) that runs natively, works offline, and keeps your DNA private.
GenomeForge brings the kind of tooling that used to live behind cloud platforms and paywalls into a single fast, native app that runs entirely on your computer. Load a 23andMe export or a VCF and start exploring in seconds. Annotate variants against the major public databases, jump straight to well-known health SNPs, read the primary literature, then step next door to design a plasmid or a CRISPR edit. No account, no upload, no subscription.
- Private by design. Your raw genotypes never leave the machine. Only rsIDs and genomic regions are ever sent to public APIs, and an offline mode uses nothing but the local cache.
- Native, not a web app. Pure Rust and egui: a small, quick binary with no browser engine, no Electron, and no telemetry.
- Six tools in one window. A genome browser, a plasmid CAD, a CRISPR studio, a polygenic-risk modeler, an ancestry & haplogroup explorer, and a pharmacogenomics report, switchable from the top bar. No more stitching together half a dozen sites.
- Approachable and professional. A single dropdown retunes the entire interface from high-school-friendly to expert-dense, with a built-in tutorial in every section.
- Free and open source. MIT-licensed and multiplatform: Windows, Linux, and macOS.
Research and educational use only. Not medical advice.
Six workspaces share one window, switchable from the top bar.
| Workspace | What it does |
|---|---|
| Genome browser & variant explorer | Import 23andMe / AncestryDNA, VCF, FASTA, or GenBank. Variants are drawn as lollipops coloured by ClinVar significance and annotated against ClinVar, gnomAD, and dbSNP (via MyVariant.info). One-click quick-jumps to common health SNPs, a filterable variant table you can export to CSV, GRCh37 ⇄ GRCh38 liftover, and a per-variant deep dive with links to dbSNP, ClinVar, GWAS Catalog, gnomAD, MedlinePlus, and a live PubMed search. |
| Plasmid designer | Circular and linear maps, restriction-site mapping with unique-cutter highlighting, ORF detection and translation, primer Tm / GC, a digest / cloning preview with a mini gel, and Golden-Gate / Gibson assembly simulation. |
| CRISPR studio | Guide design across SpCas9, SpCas9-NG, SpRY, and Cas12a, with a real Doench 2014 on-target model, CFD off-target scoring (against the target or a loaded reference genome), NHEJ / HDR edit simulation, base- and prime-editing previews, and an AAV cargo planner. |
| Phenotype & risk | Polygenic scores from the PGS Catalog applied to your genome, with coverage, an ancestry caveat, and the Mendelian / ClinVar findings that can matter on their own. |
| Ancestry & lineage | Predicted maternal (mtDNA) and paternal (Y-DNA) haplogroups from PhyloTree and the ISOGG tree, plus a continental ancestry-composition estimate against five 1000 Genomes super-populations. Computed entirely on your machine, framed as coarse genetic similarity rather than an identity test. |
| Pharmacogenomics | A CPIC-based drug-response report: star-allele / diplotype calling for 15 pharmacogenes with metabolizer phenotypes and per-drug notes, honest about what consumer arrays cannot call. Informational only, not medical advice. |
Open files by drag-and-drop or File ▸ Open Recent, copy sequences and IDs to the clipboard, and let the app check for new releases on launch.
A dropdown in every workspace retunes the whole interface, and each section has a short built-in tutorial:
- Beginner: plain language and guided, written for a student or the simply curious.
- Intermediate: the same, plus the numbers and what they mean.
- Expert: dense and complete, for biologists and genomics professionals.
Grab the latest build for your platform from the Releases page:
| Platform | Files |
|---|---|
| Windows | GenomeForge-<version>-x64.msi (installer) or the portable .exe |
| Linux | .tar.gz (portable binary) or a Flatpak bundle |
| macOS | universal .dmg (Intel + Apple Silicon) |
No genome of your own? The app ships with a synthetic demo: File ▸ Open ▸ assets/sample_genome_23andme.txt (or just drag it onto the window).
The builds aren't yet code-signed, so each OS will warn you the first time. This is expected; here's how to get past it:
- Windows: SmartScreen may show "Windows protected your PC." Click More info ▸ Run anyway. (The MSI installs per-user; no admin needed.)
- macOS: Gatekeeper will say the app "can't be opened." Right-click the app ▸ Open ▸ Open, or run
xattr -dr com.apple.quarantine /Applications/GenomeForge.app. The build is universal (Intel + Apple Silicon). - Linux: the tarball binary just runs (
./GenomeForge);chmod +xit if needed. For the Flatpak:flatpak install --user GenomeForge-*.flatpakthenflatpak run io.github.abowlofeleven.GenomeForge.
Signing/notarization is on the roadmap; until then, the source is right here if you'd rather build it yourself.
GenomeForge is a pure-Rust workspace. With a recent stable toolchain:
cargo run -p gx-app # launch
cargo run -p gx-app -- assets/sample_genome_23andme.txt # open a demo genome
cargo test --workspace # run the testsPer-platform packaging (Windows MSI, Linux Flatpak, macOS .dmg) is documented in
BUILD.md.
A Cargo workspace of small, focused crates:
| Crate | Responsibility |
|---|---|
gx-core |
Domain model: assemblies, ranges, variants, features |
gx-io |
Importers: raw DNA, VCF, FASTA / GenBank |
gx-annotate |
Hybrid online + cached annotation (MyVariant, Ensembl, PGS Catalog, PubMed) over a local cache |
gx-plasmid |
Restriction mapping, ORFs, primers, digest / cloning |
gx-crispr |
Guide design, Doench / CFD scoring, edit simulation, AAV planning |
gx-pgs |
PGS Catalog parsing and scoring |
gx-app |
The eframe (egui) desktop application |
Built with egui / eframe, noodles, gb-io, and redb. Every network and disk operation runs on a background thread, so the interface never blocks.
- Raw genotypes stay on your machine; only rsIDs and regions are sent to annotation APIs.
- An offline toggle disables all network access and renders from the local cache.
- Annotations are cached locally for fast, offline reuse.
MIT. © GenomeForge contributors.
Scoring methods (Doench on-target, CFD off-target, PGS percentiles) are transparent, documented decision-support estimates, not a clinical pipeline. GenomeForge is for learning, research, and design exploration, not medical advice.