Every screened shotgun-metagenome project in ENA, NCBI SRA and DDBJ (64,369), classified for human host, body site, life stage and assay; ✓ marks the projects in the catalog.
About this page
Is the registry all human? No. The registry holds every shotgun-metagenome study reached by the enumeration — INSDC BioProjects deposited in ENA, NCBI SRA or DDBJ and enumerated through ENA's mirror — and classifies each for human host (registry_studies.host_human): 6,652 host human = yes, 246 mixed (human and non-human material), 13,696 unknown and 43,775 no, out of 64,369 studies in release R2026.25. The 6,898 human studies (yes + mixed) are the headline number on this site; the explorer below shows them by default and lets you include the unknown and non-human rows.
Registry explorer
Filters the 64,369 registry studies in your browser — by default the 6,898 human studies (host human = yes or mixed); tick unknown / no to widen. Runs using DuckDB-WASM over data/registry_studies.parquet (9.0 MB); nothing is sent to a server. Click a row for every column, the evidence rows behind host / body site / life stage. The URL updates with your filters so you can share a view. Registry-only studies have no page on this site (they link to ENA); catalog studies link to their study page. Opening registry/index.html?study=PRJ… filters to that study and opens its record.
✓ = in the curated catalog (has a project page with per-sample metadata). Click any row to see why a study is — or is not — in the catalog.
Loading DuckDB-WASM (from cdn.jsdelivr.net, pinned 1.29.0) and the registry table…
…
Columns shown are a subset; the CSV export carries all 51 columns of the filtered rows. Full table: registry_studies.parquet. Per-slice enumeration completeness: registry_universe_audit.csv. The run-level table (registry_runs_v1.20.0.parquet, 2,334,512 rows, ≈70 MB, 46 ENA fields + found_by) is attached to the GitHub Release of the data repository, not a site file.
Scopes
Scope memberships derive deterministically from (host, body sites, life stages, assay) — see config/scope.yaml. A study can belong to several scopes. Click a scope for its definition, counts and largest studies.
Scope membership (a study counts once per scope)(scope_memberships)
value
studies
runs
human_allAll human shotgun metagenomes (registry)
4,675
929,502
gut_adultAdult gut (≥ 18 y)
1,389
369,164
blood_tissueBlood / tissue
1,251
161,229
unknown_siteBody site unknown
828
57,717
respiratoryRespiratory tract
827
117,408
oralOral cavity
613
140,572
gut_childChild / adolescent gut (3–18 y)
406
147,159
infant_gutInfant gut (0–3 y)
389
182,725
vaginal_urogenitalVaginal / urogenital
289
67,023
other_siteOther body site
278
45,273
skinSkin
273
77,062
milkHuman milk
64
22,569
Classification stage(classification_stage)
value
studies
runs
deterministic_ruleENA-field rule (host taxon 9606, library fields, numeric age with unit)
52,142
862,817
sonnet_x2two replicate model classifications in agreement
5,164
643,443
deterministic_priorprior carried over from an earlier triage (reason code / body-site call) or deterministic term match
4,808
524,690
opus_adjudicatedreplicate disagreement adjudicated by the stronger model
1,663
243,379
pendingnot yet classified
417
21,564
owner_decisionstudy-level decision by the catalog owner (config/registry_overrides.yaml)
96
20,336
curator_auditstudy-level host/assay audit of the archive records (config/registry_overrides.yaml)
79
18,283
Host human(host_human)
value
studies
runs
no
43,775
819,394
unknown
13,696
343,492
yes
6,652
1,118,016
mixed
246
53,610
Access(access)
value
studies
runs
open
64,368
2,321,267
controlled
1
13,245
What a registry row is
Each study carries archive-only metadata (route R1): host, body site(s) anchored to UBERON, life stage(s), assay, access, and the classification stage that produced them, each committed value with an evidence row (labelled source, ≤ 12-word quote). Registry rows are classifications, not inclusions: a study can carry several body sites and life stages; rows at stage pending have not been classified yet. The curated tier goes deeper — per-sample fields recovered from supplementary tables and papers (routes R2–R4) — and is built scope by scope; the catalog (human gut, all ages: 2,798 studies with study pages) is the first curated scope. See Methods › Registry classification for the cascade, the vocabularies and what the confidence means.
6,898
human studies (host human = yes 6,652 + mixed 246), 1,171,626 runs · 64,369 INSDC BioProjects screened in total
2,334,512
sequencing runs
637,397
BioSamples with harvested attributes (registry_biosamples.parquet, 4,381 studies): body site known for 51 %, age 16 %, sex 19 %
838,702
runs with a Sandpiper community profile (36 % of runs; 37,328 studies)
2,027,594
BioSamples
43,775
non-human (host human = no) · 13,696 unknown — screened and kept for provenance, excluded from the default explorer view
63,952
classified (99%) · 417 pending
12
body-site scopes; 2,798 studies are in the curated catalog (human gut, all ages)