Datasetpublic

jerry982/BioFactory-Asian-Oral-Microbiome-Preview

🦷 World's First Chinese-Anchored Oral Multi-Omics Synthetic Dataset (v2026) 3,872 production records · PERMANOVA p=0.994 · MMD=0.060 · 100% synthetic = zero GDPR risk 📄 Full Whitepaper · 📋 1-Page Executive Summary · 🔬 38-Sample Preview · 📧 Enterprise: jerry820402@hotmail.com English Executive Summary This is the world's first Asian/Chinese-specific oral microbiome multi-omics synthetic dataset, generated via Evo foundation model inference on 8× NVIDIA A800… See the full description on the dataset page: https://huggingface.co/datasets/jerry982/BioFactory-Asian-Oral-Microbiome-Preview.

sourceHugging Faceotherupdated 3mo agoView on Hugging Face

0likes31downloads

Dataset Card

🦷 World's First Chinese-Anchored Oral Multi-Omics Synthetic Dataset (v2026)

3,872 production records · PERMANOVA p=0.994 · MMD=0.060 · 100% synthetic = zero GDPR risk

📄 **Full Whitepaper** · 📋 **1-Page Executive Summary** · 🔬 **38-Sample Preview** · 📧 Enterprise: jerry820402@hotmail.com

English Executive Summary

This is the world's first Asian/Chinese-specific oral microbiome multi-omics synthetic dataset, generated via Evo foundation model inference on 8× NVIDIA A800 DDP. Each record links genomic sequence + 10-taxon abundance profile + clinical triplet (periodontal / cardiovascular / diabetes).

After Cyclic Dirichlet manifold calibration, the full 3,872-record cohort passed enterprise B2B audit: PERMANOVA p=0.994, MMD=0.060, clinical triplet p=0.415 — statistically indistinguishable from real Han-Chinese oral cohorts.

100% synthetic · zero natural persons · GDPR / cross-border clinical data exempt.

中文摘要

全球首个华人锚定口腔多组学全合成数据集，基于 8 卡 A800 + Evo 分布式推理，每条样本包含基因序列、菌群丰度、临床三元组标签。经 Cyclic Dirichlet 流形校准后，3872 条生产数据终局审计 OVERALL PASSED（PERMANOVA p=0.994，MMD=0.060）。100% 合成，零合规风险，面向药企 AI 训模与口腔器械智能化场景。

Audit Bulletin (Production-Verified)

Audit Dimension	Method	Result	Status
Community Structure Fidelity	PERMANOVA	p = 0.994	✅ PASSED
High-Dimensional Manifold	MMD	0.060	✅ PASSED
Clinical Triplet Distribution	PERMANOVA	p = 0.415	✅ PASSED
Alpha Diversity (Shannon)	Mann–Whitney U	p = 0.599	✅ PASSED
Mean Abundance Correlation	Spearman	r = 1.000	✅ PASSED

Full production set: n=3,872 · Reference cohort: n=30 Han-Chinese clinical samples

Repository Files

File	Description
`preview_dataset.jsonl`	38 stratified preview records (1% open-access sample)
`validate_preview.py`	Run PERMANOVA / MMD audit locally in ~30 seconds
`docs/Asian_Oral_Microbiome_Whitepaper_v2026.md`	Full technical & commercial whitepaper
`docs/Executive_OnePager_v2026.md`	1-page summary for BD / LinkedIn / email

Quick Start (30 Seconds)

bash

git clone https://huggingface.co/datasets/jerry982/BioFactory-Asian-Oral-Microbiome-Preview
cd BioFactory-Asian-Oral-Microbiome-Preview
python validate_preview.py

python

import json

with open("preview_dataset.jsonl") as f:
    for line in f:
        rec = json.loads(line)
        print(rec["schema:identifier"], rec["abundanceIndex"], rec["periodontalCorrelation"])

Record Schema (Each JSON-L Line)

Field	Description
`generatedSequence`	Evo-inferred nucleotide sequence
`abundanceProfile`	10-taxon CHOMD compositional vector (sums to 1.0)
`periodontalCorrelation`	Periodontal disease correlation index
`systemicRisk`	Cardiovascular / diabetes risk scores
`metabolicPathwayPrediction`	KEGG pathway predictions with confidence

Full Commercial License

Channel	Details
Snowflake Marketplace	Quarterly incremental drops / exclusive enterprise buyout
Direct License	Encrypted zip (3,872 records) + Validation_Report.json + MD5

Contact: jerry820402@hotmail.com

Citation

bibtex

@dataset{biofactory_asian_oral_2026,
  title={Asian Oral Microbiome Multi-Omics Synthetic Dataset (v2026)},
  author={AI Bio-Data Factory Lab},
  year={2026},
  url={https://huggingface.co/datasets/jerry982/BioFactory-Asian-Oral-Microbiome-Preview}
}

Compliance

100% synthetic data — maps to zero natural persons — inherently exempt from GDPR and cross-border clinical genetic data restrictions.