About
The African DSI DataBank is a federated, hub-and-spoke platform for hosting, mirroring, curating, and governing access to Africa-origin biodiversity and agricultural Digital Sequence Information (DSI): genomes and transcriptomes, protein sequences and experimental proteomics, predicted protein structures, eDNA/metabarcoding and metagenomics, chemical (targeted) and metabolic (untargeted) profiles, and field/specimen and cellular/molecular imaging. Data flows upward from institution-owned Spoke nodes through National and Sub-regional Hubs to one Continental Hub, with full provenance preserved at every step, and can flow back down as real bytes — not just metadata — through a consent-gated transfer mechanism that always requires both sides to agree.
Capability maturity
A three-way split of what's actually running versus what's a placeholder. "Verified live" means actually run against a real external service, a real browser, or a real multi-service deployment — not just covered by the automated test suite.
| Capability | Status |
|---|---|
| Federated sync chain (Spoke → National → Sub-regional → Hub) | ✅ Verified live |
| Central JWT authentication, roles, node admission/approval | ✅ Verified live |
| Node<->Hub bidirectional data transfer (genomes, transcriptomes, proteins, eDNA/metagenomic/proteomics/metabolomics raw samples) | ✅ Verified live |
| Third-party node backend support (documented, versioned wire protocol + automatic capability handshake) | ✅ Verified live — verified against both a real stock node and a real, independently-written third-party node |
| Genome browser (JBrowse2) | ✅ Verified live — pixel-confirmed via headless-browser session |
| NCBI/ENA genome mirroring, GoaT assembly-quality enrichment, GBIF image mirroring | ✅ Verified live — against the real external APIs |
| Cellular/molecular imaging metadata lookup (EMPIAR-shaped and BioImage-Archive-shaped sources) | ✅ Verified live — against the real external APIs |
| Production Docker deployment (Postgres/Redis/MinIO, non-root containers, Caddy rate limiting) | ✅ Verified live — built and run for real, including a live rate-limit flood test |
| Native (non-Docker) install — systemd units + .deb/.rpm packaging, Hub and node tiers | ✅ Verified live — verified on real Debian- and RHEL-family systemd hosts |
| API versioning (/api/v1) with real pagination on every Hub list endpoint | ✅ Verified live |
| Per-species data-completeness dashboard | ✅ Verified live |
| Faceted genome search (country, license, origin, assembly quality, taxonomic group/kingdom, A-Z browsing) | ✅ Verified live |
| Genome-triggered auto-sync of related data from external sources (UniProt, PDBe, AlphaFold DB, PaxDb, PRIDE, ANPDB, ChEMBL, GBIF, iDigBio, BioImage Archive, IDR, MetaboLights, KEGG) | ✅ Verified live — each source verified against its own real, live API before shipping; CIL specifically is a real, tested client but currently returns zero results due to a confirmed cert-chain misconfiguration on CIL's own server, not a bug in this code |
| Baobab Index — real bibliographic metadata mirrored from PubMed | ✅ Verified live — verified against the real, live PubMed API, including a real PMC-to-PMID resolution path |
| Multi-target compute search (BLAST similarity search, phylogenetic tree build, containerized bioinformatics tools) across the Hub and any number of connected nodes at once | ✅ Verified live — structure prediction (ESMFold) has its own separate, dedicated flow, not part of this general dispatch |
| Support ticket system (submission, threaded replies, admin queue) | ✅ Verified live |
| Ticket sync to an external tracker (GitHub Issues, Linear, Jira) | 🟡 Implemented, unverified — built and tested against each service's own documented API/webhook contract — not yet exercised against a real repo/workspace/PAT |
| CAPTCHA (Cloudflare Turnstile) on public write endpoints | 🟡 Implemented, unverified — built and tested against Cloudflare's own documented test keys — no real production Turnstile account registered yet |
| Continental-Admin-controlled nav visibility (feature flags, both frontends) | ✅ Verified live |
| Bulk export (Darwin Core Archive / DCAT catalog) | ✅ Verified live — meta.xml/eml.xml structure verified against GBIF's own Darwin Core Text Guidelines |
| Structure viewer (Mol*) | 🟡 Implemented, unverified — not yet visually confirmed live — no browser available at build time |
| DDBJ genome mirroring (file fetch specifically) | 🟡 Implemented, unverified — matches DDBJ's documented API shape, not confirmed against the live service |
| Third-party ABS (access-and-benefit-sharing) identity verification | 🟡 Implemented, unverified — built against a documented contract — no live third-party ABS service exists yet |
| Real DOI minting | 🟡 Implemented, unverified — scaffolded, inert until a DataCite prefix is registered |
| Transactional email (password reset, Contact Us) | 🟡 Implemented, unverified — two real implementations — a dedicated API (Resend) and a real SMTP mailbox — neither yet exercised against a live account/mailbox in this repo's own testing |
| CBD provenance / country-of-origin / content-checksum fields (Increment 26) | 🟡 Implemented, unverified — fully implemented and tested, not yet exercised against a live multi-node deployment |
| Protein domain annotation (Pfam/InterPro via EBI InterProScan5) | 🟡 Implemented, unverified — submit + status legs confirmed live; the result-fetch leg wasn't reached before a live job finished |
| Structure prediction (ESMFold) | 🔴 Pluggable stub — proves the job/versioning pipeline only |
| eDNA classification (DADA2/QIIME2) | 🔴 Pluggable stub |
| Metagenomics profiling (Kraken2/HUMAnN) | 🔴 Pluggable stub |
| Proteomics / metabolomics sample analysis (mass-spec/NMR) | 🔴 Pluggable stub — the raw-sample-plus-derived-result sync/transfer pipeline itself is real and live-verified; only the analysis step is a stub, same as eDNA/metagenomics |
| Species/image ML classification, specimen-label OCR | 🔴 Pluggable stub |
| ENA submission — Study/Sample registration, Webin-CLI assembly submission, accession polling, and a production-cutover switch (Phases 2-5) | 🟡 Implemented, unverified — written to ENA's own documented sandbox/Webin-CLI/Portal Search API contracts — real Study/Sample registration, a real Webin-CLI validate/submit wrapper, and real accession polling (against ENA's public, keyless Portal Search API, correlated on the record's own deterministic assembly name) plus a Continental-Admin review UI, fully test-covered; a real, sha256-verified Webin-CLI JAR is bundled into the Hub's Docker image and native install. The whole pipeline sits behind a platform-wide Continental-Admin toggle (default off — no submission control is even shown until it's turned on), and once ENA accepts an accession a further, independent sweep confirms it has genuinely propagated through INSDC's own documented daily NCBI/ENA/DDBJ synchronisation, not just that ENA itself accepted it. A settings-gated production-cutover switch exists — only the exact literal "production" ever targets ENA's real service, everything defaults to sandbox — but not yet exercised against a live account (no ENA Webin credentials registered yet). |
| Tarball-based node software distribution (signed + checksummed releases, no git access needed) | ✅ Verified live — verified live end-to-end — a real GPG release-signing keypair, a real published release, and a real first-install simulation from a genuinely separate, freshly-hardened server with zero prior git access to this repo: apply for an account, get approved, log in, download the gated tarball, independently cross-check the signing key's fingerprint, then verify both the manifest's signature and the artifact's checksum. |
| Embargoed access tier — auto-release on a set date | ✅ Verified live — a fourth access_tier alongside public/regional_researcher/restricted, with two submitter-chosen sub-levels during the embargo window (restricted-like or regional-researcher-like) and a real scheduled release sweep (15 min) plus a request-time lazy-release backstop — now available across all 11 citable record types (genomes, proteins, structures, eDNA classifications, metagenomic/proteomics/metabolomics profiles, field/specimen and cellular/molecular imaging, and both proteomics/metabolic project references), not just genomes. |
Governance & policy
Governance (node admission, Continental override, dispute resolution), Terms of use & citation policy (data use and citation policy), and Preservation & sustainability (preservation and funding model).
Development & attribution
Copyright in this platform is held by the African BioGenome Project (AfricaBP). It was designed and built by ThankGod Echezona Ebenezer, implementing the roadmap AfricaBP published in Nature Reviews Biodiversity (Katee, Botes, Hayah et al., 2026) — ThankGod Echezona Ebenezer is a co-author of that paper. See this repository's own CITATION.cff for the machine-readable, citable software-authorship record.