Home › Collections
Collections (32)
A collection is a named, curated set of catalog studies or samples with its filters pre-entered: the reference cohorts everyone re-analyses, landmark populations, disease reference sets, dense longitudinal designs, age bands.
About this page
Every count below is computed at build time on the released sample table, so it equals what the sample explorer shows when you open the collection. Membership rules live in config/collections.yaml; to propose a collection or a correction, open a pull request.
Benchmark and reference cohorts (5)
HMP1 healthy adult reference (Human Microbiome Project, WGS)
The NIH Human Microbiome Project phase 1 shotgun metagenomes of clinically healthy US adults (Houston and St Louis), the most reused healthy-adult reference set; the BioProject spans all HMP body sites, so restrict to gut samples with body_site_class.
2 studies · 530 samples HMP2 / IBDMDB longitudinal IBD multi-omics cohort
The Integrative HMP inflammatory bowel disease cohort (IBDMDB): 132 subjects with Crohn's disease, ulcerative colitis or no IBD, followed for one year with stool metagenomes, metatranscriptomes and biopsies at five US centres.
2 studies · 2,341 samples MetaHIT European gut metagenome cohorts
The MetaHIT consortium deposits underlying the first gut gene catalogue and the metagenomic-species work: Danish and Spanish adults (Qin 2010), Danish obese/lean adults (Le Chatelier 2013, Nielsen 2014) and the integrated gene catalogue samples (Li 2014).
4 studies · 1,072 samples Curated healthy adult gut samples (all studies)
Every gut sample in the curated catalog whose source states the subject is healthy or a control and whose age at collection falls in the adult band (18–65 y). The catalog's own counterpart of the curatedMetagenomicData healthy-adult set; membership is evidence-backed per sample.
540 studies · 57,647 samples Protocol benchmarks with mock-community controls
Human stool studies that sequenced a defined mock community (ZymoBIOMICS or in-house standards) alongside real samples to compare DNA extraction kits, library protocols or 16S versus shotgun profiling — useful as positive controls and for bias estimation.
5 studies · 577 samples
The NIH Human Microbiome Project phase 1 shotgun metagenomes of clinically healthy US adults (Houston and St Louis), the most reused healthy-adult reference set; the BioProject spans all HMP body sites, so restrict to gut samples with body_site_class.
2 studies · 530 samples HMP2 / IBDMDB longitudinal IBD multi-omics cohort
The Integrative HMP inflammatory bowel disease cohort (IBDMDB): 132 subjects with Crohn's disease, ulcerative colitis or no IBD, followed for one year with stool metagenomes, metatranscriptomes and biopsies at five US centres.
2 studies · 2,341 samples MetaHIT European gut metagenome cohorts
The MetaHIT consortium deposits underlying the first gut gene catalogue and the metagenomic-species work: Danish and Spanish adults (Qin 2010), Danish obese/lean adults (Le Chatelier 2013, Nielsen 2014) and the integrated gene catalogue samples (Li 2014).
4 studies · 1,072 samples Curated healthy adult gut samples (all studies)
Every gut sample in the curated catalog whose source states the subject is healthy or a control and whose age at collection falls in the adult band (18–65 y). The catalog's own counterpart of the curatedMetagenomicData healthy-adult set; membership is evidence-backed per sample.
540 studies · 57,647 samples Protocol benchmarks with mock-community controls
Human stool studies that sequenced a defined mock community (ZymoBIOMICS or in-house standards) alongside real samples to compare DNA extraction kits, library protocols or 16S versus shotgun profiling — useful as positive controls and for bias estimation.
5 studies · 577 samples
Population cohorts (5)
American Gut Project shotgun subset
Shotgun metagenomes from the crowd-sourced American Gut Project (US, UK and Australia self-collected stool with extensive diet and lifestyle questionnaires); the largest single citizen-science deposit in the catalog.
1 studies · 7,353 samples PREDICT personalised-nutrition cohorts (UK/US)
ZOE PREDICT 1, 2 and 3 deep-phenotyped adults (TwinsUK and volunteers in the UK and US) with habitual-diet records and postprandial metabolic responses; PREDICT 3 adds more than 20,000 stool metagenomes.
4 studies · 22,895 samples Hadza hunter-gatherers (Tanzania)
All shotgun deposits from the Hadza of Tanzania, the reference hunter-gatherer population: Rampelli 2015, the seasonal-cycling cohort (Smits 2017) and the deep-sequenced adults and infants of Carter 2023 / Olm 2022 (with Nepali comparison groups in the same BioProject).
1 studies · 40 samples Amazonian and Amerindian communities
Indigenous South American communities with traditional subsistence: Matses hunter-gatherers and Tunapuco highland agriculturalists (Peru), Yanomami and Yekwana villages (Venezuela, Brazil) and the Wayampi of French Guiana — the low-industrialization reference for Prevotella/Treponema-rich microbiomes.
1 studies · 15 samples Controlled-access population cohorts (reference only)
Large general-population gut metagenome cohorts whose raw reads are deposited under controlled access and therefore cannot be curated here: LifeLines-DEEP (Netherlands, Zhernakova et al. 2016) has only its metagenome-assembled genomes in the public archive (PRJEB76860, 0 read runs). Listed so the gap is explicit; the collection carries no samples.
0 studies · 0 samples
Shotgun metagenomes from the crowd-sourced American Gut Project (US, UK and Australia self-collected stool with extensive diet and lifestyle questionnaires); the largest single citizen-science deposit in the catalog.
1 studies · 7,353 samples PREDICT personalised-nutrition cohorts (UK/US)
ZOE PREDICT 1, 2 and 3 deep-phenotyped adults (TwinsUK and volunteers in the UK and US) with habitual-diet records and postprandial metabolic responses; PREDICT 3 adds more than 20,000 stool metagenomes.
4 studies · 22,895 samples Hadza hunter-gatherers (Tanzania)
All shotgun deposits from the Hadza of Tanzania, the reference hunter-gatherer population: Rampelli 2015, the seasonal-cycling cohort (Smits 2017) and the deep-sequenced adults and infants of Carter 2023 / Olm 2022 (with Nepali comparison groups in the same BioProject).
1 studies · 40 samples Amazonian and Amerindian communities
Indigenous South American communities with traditional subsistence: Matses hunter-gatherers and Tunapuco highland agriculturalists (Peru), Yanomami and Yekwana villages (Venezuela, Brazil) and the Wayampi of French Guiana — the low-industrialization reference for Prevotella/Treponema-rich microbiomes.
1 studies · 15 samples Controlled-access population cohorts (reference only)
Large general-population gut metagenome cohorts whose raw reads are deposited under controlled access and therefore cannot be curated here: LifeLines-DEEP (Netherlands, Zhernakova et al. 2016) has only its metagenome-assembled genomes in the public archive (PRJEB76860, 0 read runs). Listed so the gap is explicit; the collection carries no samples.
0 studies · 0 samples
Lifestyle (5)
Olm 2022 Hadza infant cohort (lifestyle spectrum)
Infant stool metagenomes from the Hadza deposited with Olm et al. 2022 (Science), the non-industrialized anchor of the infant lifestyle-spectrum comparison; restricted to the neonate/infant age bands of BioProject PRJEB49206.
1 studies · 121 samples Industrialised vs non-industrialised comparisons
Cohorts designed around an urbanisation or industrialisation contrast: Cameroon and African rural–urban gradients, Ghanaian and Malagasy villages (Pasolli 2019), isolated Honduran villages, Fijian villages, Orang Asli, the Baiga tribe and Irish Travellers.
3 studies · 2,651 samples High-altitude and plateau populations
Residents of high-altitude environments: Tibetan and Yi populations of the Qinghai–Tibet plateau, Papua New Guinea highlanders and a mountaineering-expedition cohort.
2 studies · 75 samples Athletes and elite sport
Cohorts defined by sport: professional and elite cyclists, Irish and Chinese elite athletes, marathon runners, recreational athletes stratified by fitness, and sport-nutrition trials.
5 studies · 274 samples Vegan and vegetarian diet cohorts
Cohorts recruited by habitual diet — vegans, vegetarians and omnivore comparison groups — including the Stanford identical-twin vegan trial and Central European vegan/omnivore cohorts.
0 studies · 0 samples
Infant stool metagenomes from the Hadza deposited with Olm et al. 2022 (Science), the non-industrialized anchor of the infant lifestyle-spectrum comparison; restricted to the neonate/infant age bands of BioProject PRJEB49206.
1 studies · 121 samples Industrialised vs non-industrialised comparisons
Cohorts designed around an urbanisation or industrialisation contrast: Cameroon and African rural–urban gradients, Ghanaian and Malagasy villages (Pasolli 2019), isolated Honduran villages, Fijian villages, Orang Asli, the Baiga tribe and Irish Travellers.
3 studies · 2,651 samples High-altitude and plateau populations
Residents of high-altitude environments: Tibetan and Yi populations of the Qinghai–Tibet plateau, Papua New Guinea highlanders and a mountaineering-expedition cohort.
2 studies · 75 samples Athletes and elite sport
Cohorts defined by sport: professional and elite cyclists, Irish and Chinese elite athletes, marathon runners, recreational athletes stratified by fitness, and sport-nutrition trials.
5 studies · 274 samples Vegan and vegetarian diet cohorts
Cohorts recruited by habitual diet — vegans, vegetarians and omnivore comparison groups — including the Stanford identical-twin vegan trial and Central European vegan/omnivore cohorts.
0 studies · 0 samples
Disease reference sets (8)
Colorectal cancer meta-analysis cohorts (Wirbel / Thomas 2019)
The eight case–control shotgun cohorts pooled by Wirbel et al. and Thomas et al. (Nature Medicine 2019): France, Austria, Germany, China, USA, two Italian cohorts and Japan — the standard benchmark for cross-cohort CRC classifiers.
8 studies · 1,641 samples All colorectal cancer and adenoma samples
Every gut sample in the catalog whose evidence assigns colorectal cancer or colorectal adenoma/polyp as the health condition, whatever the study — for extending the CRC benchmark beyond the 2019 meta-analysis cohorts.
85 studies · 6,334 samples Type 2 diabetes reference cohorts (Qin 2012, Karlsson 2013)
The two founding T2D metagenome-wide association cohorts — Chinese adults (Qin 2012) and 70-year-old European women with normal, impaired and diabetic glucose control (Karlsson 2013) — plus the large Gothenburg pre-diabetes/T2D cohort.
3 studies · 2,341 samples All type 2 diabetes / prediabetes samples
Every gut sample assigned type 2 diabetes or prediabetes by the curated health_condition, across all studies.
66 studies · 5,177 samples IBD reference cohorts
Landmark inflammatory bowel disease shotgun cohorts: HMP2/IBDMDB, the PRISM cohort (Franzosa 2019), the Broad longitudinal IBD stool study (Hall 2017) and the disease-adapted-lineage cohort of 2024.
5 studies · 3,783 samples All IBD samples (Crohn's, ulcerative colitis, unspecified)
Every gut sample whose curated health_condition is Crohn's disease, ulcerative colitis or IBD unspecified, across all studies.
169 studies · 15,616 samples Liver cirrhosis cohorts
Shotgun cohorts of liver cirrhosis: the Chinese case–control cohort of Qin 2014 (the cirrhosis MWAS), the MICROB-PREDICT decompensated-cirrhosis deposit and the PROFIT FMT-in-cirrhosis trial. Pair with the liver_disease health-condition filter for all liver-disease samples.
3 studies · 892 samples All liver disease samples
Every gut sample whose curated health_condition is liver disease (cirrhosis, hepatitis, hepatocellular carcinoma, hepatic encephalopathy), across all studies.
50 studies · 5,480 samples
The eight case–control shotgun cohorts pooled by Wirbel et al. and Thomas et al. (Nature Medicine 2019): France, Austria, Germany, China, USA, two Italian cohorts and Japan — the standard benchmark for cross-cohort CRC classifiers.
8 studies · 1,641 samples All colorectal cancer and adenoma samples
Every gut sample in the catalog whose evidence assigns colorectal cancer or colorectal adenoma/polyp as the health condition, whatever the study — for extending the CRC benchmark beyond the 2019 meta-analysis cohorts.
85 studies · 6,334 samples Type 2 diabetes reference cohorts (Qin 2012, Karlsson 2013)
The two founding T2D metagenome-wide association cohorts — Chinese adults (Qin 2012) and 70-year-old European women with normal, impaired and diabetic glucose control (Karlsson 2013) — plus the large Gothenburg pre-diabetes/T2D cohort.
3 studies · 2,341 samples All type 2 diabetes / prediabetes samples
Every gut sample assigned type 2 diabetes or prediabetes by the curated health_condition, across all studies.
66 studies · 5,177 samples IBD reference cohorts
Landmark inflammatory bowel disease shotgun cohorts: HMP2/IBDMDB, the PRISM cohort (Franzosa 2019), the Broad longitudinal IBD stool study (Hall 2017) and the disease-adapted-lineage cohort of 2024.
5 studies · 3,783 samples All IBD samples (Crohn's, ulcerative colitis, unspecified)
Every gut sample whose curated health_condition is Crohn's disease, ulcerative colitis or IBD unspecified, across all studies.
169 studies · 15,616 samples Liver cirrhosis cohorts
Shotgun cohorts of liver cirrhosis: the Chinese case–control cohort of Qin 2014 (the cirrhosis MWAS), the MICROB-PREDICT decompensated-cirrhosis deposit and the PROFIT FMT-in-cirrhosis trial. Pair with the liver_disease health-condition filter for all liver-disease samples.
3 studies · 892 samples All liver disease samples
Every gut sample whose curated health_condition is liver disease (cirrhosis, hepatitis, hepatocellular carcinoma, hepatic encephalopathy), across all studies.
50 studies · 5,480 samples
Study design (6)
Twin and family cohorts
Twin-pair and family designs that separate host genetics from shared environment: TwinsUK shotgun deposits, Turnbaugh 2009 obese/lean twins, Korean and Chinese twins, Malawian twins discordant for malnutrition, and twin viromes.
10 studies · 3,848 samples Densely sampled subjects (≥ 5 gut samples per subject)
Every subject in the catalog with at least five gut samples linked to the same subject_id — the time-series slice for strain tracking, stability and perturbation-recovery analyses. Dominated by infant birth cohorts (TEDDY, DIABIMMUNE), IBDMDB and adult time-series studies.
444 studies · 102,587 samples Adult time-series reference cohorts
Healthy-adult cohorts sampled repeatedly over months to years: the Broad/OpenBiome BIO-ML donors, dense FMT-donor series, the Gothenburg longitudinal non-disease cohort, HPFS men sampled twice six months apart and the ciprofloxacin time-series volunteers.
5 studies · 3,763 samples Faecal microbiota transplantation trials
Randomised and single-arm FMT trials with shotgun follow-up across indications — ulcerative colitis, IBS, obesity, Parkinson's disease, hypertension, cirrhosis, diabetic neuropathy, immunotherapy resistance in melanoma and lung cancer, and CPE decolonisation — plus FMT donor series.
19 studies · 5,376 samples Antibiotic intervention and recovery cohorts
Studies in which antibiotics were administered by design or documented as the exposure of interest, with before/after shotgun sampling: healthy-volunteer courses (cefprozil, ciprofloxacin, beta-lactams), probiotic-after-antibiotic trials, travellers'-diarrhoea regimens, mass azithromycin distribution and preterm-infant exposure.
13 studies · 5,520 samples Samples with documented antibiotic exposure
All gut samples whose source records antibiotics at or shortly before collection (antibiotic_exposure = yes), across every study — the cross-study exposure set for resistome and recovery analyses.
193 studies · 18,320 samples
Twin-pair and family designs that separate host genetics from shared environment: TwinsUK shotgun deposits, Turnbaugh 2009 obese/lean twins, Korean and Chinese twins, Malawian twins discordant for malnutrition, and twin viromes.
10 studies · 3,848 samples Densely sampled subjects (≥ 5 gut samples per subject)
Every subject in the catalog with at least five gut samples linked to the same subject_id — the time-series slice for strain tracking, stability and perturbation-recovery analyses. Dominated by infant birth cohorts (TEDDY, DIABIMMUNE), IBDMDB and adult time-series studies.
444 studies · 102,587 samples Adult time-series reference cohorts
Healthy-adult cohorts sampled repeatedly over months to years: the Broad/OpenBiome BIO-ML donors, dense FMT-donor series, the Gothenburg longitudinal non-disease cohort, HPFS men sampled twice six months apart and the ciprofloxacin time-series volunteers.
5 studies · 3,763 samples Faecal microbiota transplantation trials
Randomised and single-arm FMT trials with shotgun follow-up across indications — ulcerative colitis, IBS, obesity, Parkinson's disease, hypertension, cirrhosis, diabetic neuropathy, immunotherapy resistance in melanoma and lung cancer, and CPE decolonisation — plus FMT donor series.
19 studies · 5,376 samples Antibiotic intervention and recovery cohorts
Studies in which antibiotics were administered by design or documented as the exposure of interest, with before/after shotgun sampling: healthy-volunteer courses (cefprozil, ciprofloxacin, beta-lactams), probiotic-after-antibiotic trials, travellers'-diarrhoea regimens, mass azithromycin distribution and preterm-infant exposure.
13 studies · 5,520 samples Samples with documented antibiotic exposure
All gut samples whose source records antibiotics at or shortly before collection (antibiotic_exposure = yes), across every study — the cross-study exposure set for resistome and recovery analyses.
193 studies · 18,320 samples
Age bands (3)
Neonates (first 28 days of life)
Every gut sample collected in the first 28 days of life (age_category = neonate), across term and preterm birth cohorts and NICU studies; the earliest colonisation window.
123 studies · 13,579 samples Centenarians and extreme longevity
Cohorts recruited for exceptional age: Japanese centenarians (Sato 2021), Sardinian semi-supercentenarians (Rampelli 2020), Chinese longevity areas (Guangxi and Sichuan cohorts), longevous families and the Greek Blue Zone cohort.
5 studies · 757 samples Older adults (≥ 65 y)
Every gut sample from subjects in the elderly age band (≥ 65 y at collection), across all studies.
379 studies · 29,302 samples
Every gut sample collected in the first 28 days of life (age_category = neonate), across term and preterm birth cohorts and NICU studies; the earliest colonisation window.
123 studies · 13,579 samples Centenarians and extreme longevity
Cohorts recruited for exceptional age: Japanese centenarians (Sato 2021), Sardinian semi-supercentenarians (Rampelli 2020), Chinese longevity areas (Guangxi and Sichuan cohorts), longevous families and the Greek Blue Zone cohort.
5 studies · 757 samples Older adults (≥ 65 y)
Every gut sample from subjects in the elderly age band (≥ 65 y at collection), across all studies.
379 studies · 29,302 samples