Simulate
Generate synthetic genomic profiles and evaluate them against the patterns and correlations relevant to a defined research objective.
Privacy-safe genomic scale for research.
Synthetic Human
Genomic Intelligence
NIOME develops AI-powered synthetic genomic data designed to preserve research-relevant patterns without relying on identifiable patient genomes.
Research Constraint
The Genomic Data Access Bottleneck
Large-scale genomic research depends on cohorts with enough diversity and statistical power to identify meaningful patterns, but access to suitable patient-level data is constrained.
Limited
Accessible patient cohorts
Research-scale
Purpose-built synthetic cohorts
Consent, governance, security, and regulatory requirements are essential safeguards, but they can make distributing identifiable genomic records slow and difficult. Research teams need alternatives that reduce direct patient-data exposure.
NIOME offers another research path.
We develop synthetic genomic profiles evaluated against research-relevant patterns and correlations without containing a direct copy of any person's genome. Scalable by design. Built to reduce patient-data exposure.
The Solution
Synthetic Genomic Intelligence at Scale
NIOME develops AI-powered synthetic genomic data for large-scale research without requiring identifiable patient genomes in downstream workflows.
Discover
Support analysis in drug response, population genetics, and precision medicine with purpose-built synthetic cohorts.
Evolve
Improve models through repeatable challenges, validator feedback, and measurable quality criteria.
Methodology & Review
01 / CHALLENGES
02 / VALIDATION
03 / ACCESS
Built to be evaluated, not just believed.
NIOME's research program is organised around explicit challenge specifications, measurable validation criteria, and defined enterprise requirements.
Defined research objectives
Each roadmap challenge sets out a biological objective, expected output, and evaluation pathway before work begins.
Review challenge specifications →Quality evaluation
Validator design considers distributional fidelity, variant relationships, reproducibility, and biological plausibility.
Read the technical whitepaper →Research-ready scoping
Enterprise cohorts are scoped around population, phenotype, format, and integration requirements before generation.
Discuss a research requirement →
Bittensor Subnet 55
Decentralized Genomic Intelligence
NIOME operates as Subnet 55 on Bittensor, where miners produce outputs and validators evaluate their quality through an incentive mechanism.
Miners
Run genomic simulation models to generate synthetic DNA profiles. Earn $TAO emissions proportional to the quality and novelty of your synthetic data.
Documentation coming soonValidators
Evaluate synthetic data quality using statistical benchmarks and biological plausibility checks. Stake $TAO to participate.
Requirements coming soonResearchers & Pharma
Scope custom cohorts around population genetics, disease variants, pharmacogenomic profiles, and delivery requirements.
Discuss Research Access →
Real-World Application
From Synthetic Data to Personalized Medicine
Here's how it works in practice: CYP2D6 is a gene that controls how your body metabolizes common drugs—from codeine to antidepressants.
NIOME challenges can ask miners to generate synthetic profiles with CYP2D6 variation, while validators evaluate the outputs against defined population and biological criteria.
The result: a privacy-conscious route to training models that study how genetic variation may affect medication response.
Miners
Synthetic Data
Validators
Verified Dataset
Researchers
Prediction Models
Enterprise Solutions
Synthetic Genomic Data for Pharmaceutical Research
Scope customised synthetic genomic datasets around your research question, population requirements, validation plan, and delivery format.
Safety
No Direct Patient RecordsGenerated cohorts are designed to avoid reproducing identifiable patient genomes and to reduce exposure of sensitive source data.
Precision
Measurable Statistical FidelityEvaluation targets population distributions, allele frequencies, and genetic relationships relevant to the research objective.
Scale
Research-Scale CohortsScope cohort size around the statistical requirements of the analysis, without distributing identifiable patient-level records.
Phenotypic Data Models
Build and train AI models that predict phenotypic outcomes from genomic data.
- Drug response prediction models
- Disease risk stratification algorithms
- Pharmacogenomic profiling systems
- Biomarker discovery pipelines
Safely Sourced Genomic Data
Our data generation pipeline is built on privacy-first principles.
- No direct patient identifiers in generated cohorts
- Population-specific validation criteria
- Privacy and governance requirements scoped per project
- Documented generation and validation workflows
In-Depth Analysis
Custom cohort generation with granular control over population characteristics.
- Ancestry-specific variant distributions
- Rare variant enrichment for edge cases
- Disease-specific genetic architectures
- Linkage disequilibrium preservation
Delivery & Integration
Plan delivery around your existing research infrastructure and analysis workflow.
- API and file-based delivery options
- VCF, PLINK, and scoped custom formats
- Batch generation requirements defined per project
- Enterprise scoping and integration support
Ready to Accelerate Your Research?
Get access to customised synthetic genomic datasets tailored to your research requirements.
Scientific Challenges
Data Challenge Roadmap
24 individually numbered research prediction challenges — building a measurable, decentralized intelligence layer for biology. Challenge 1 is complete, and challenge 2 is active.
24
Numbered Challenges
1
Complete
2
Current Challenge
22
Queued
1Challenge
CFTR Multi-Variant Haplotype
Predict protein folding stability, splicing disruption and drug response class.
2Challenge
HBB + CRISPR Editing Outcome
Predict edit success, off-target map and repair pathway (NHEJ vs HDR).
3Challenge
CYP Multi-Gene Drug Response
Predict drug metabolism rate and adverse reaction probability.
4Challenge
Synthetic Rare Disease Cohorts
Predict phenotype clusters from genotype.
5Challenge
SV + SNV Interaction
Predict phenotype from combined SNV and structural variants.
6Challenge
NGS Simulation Benchmark
Detect variants from synthetic noisy reads.
7Challenge
AAV Vector Optimization
Predict packaging efficiency and payload stability.
8Challenge
Genome-wide CRISPR Off-target
Predict off-target sites across the full genome.
9Challenge
Tissue-specific Expression
Predict expression levels across multiple tissues.
10Challenge
In Vivo Delivery Tropism
Predict vector tropism and transduction efficiency across tissues.
11Challenge
Cross-species Ortholog Modeling
Predict phenotype transferability across species.
12Challenge
Base/Prime Editing Outcome
Predict editing efficiency, bystander edits and product purity.
13Challenge
Artemisinin Pathway
Predict metabolite yield in an engineered pathway.
14Challenge
Synthetic Promoter Optimization
Predict promoter strength and expression dynamics.
15Challenge
Gene Circuit Stability
Predict circuit stability and dynamics over time.
16Challenge
Minimal Genome Viability
Predict viability of gene knockouts.
17Challenge
Protein Expression Optimization
Predict protein yield and folding efficiency.
18Challenge
Gene Circuit Evolutionary Stability
Predict long-term evolutionary robustness of circuits.
19Challenge
Vaccine Drift Simulation
Predict antigenic evolution and drift trajectory.
20Challenge
Multi-Epitope Vaccine Design
Predict immunogenicity and epitope coverage.
21Challenge
Multi-omics Integration
Predict phenotype from genome, transcriptome and proteome data.
22Challenge
Spatial Multi-Omics
Predict spatial phenotype patterns from integrated omics layers.
23Challenge
End-to-End Pipeline Simulation
Model the full pipeline from sequencing to variant calling to clinical interpretation.
24Challenge
Clinical Outcome Prediction
Predict patient-level clinical response from full pipeline outputs.
Read the full scientific foundation and challenge specifications in the NIOME Manifesto.
Backed By
Industry Leaders
This partnership is a milestone for genomics and decentralized AI.
Genomes serves as a trustless, transparent counterweight to centralized companies like 23andMe.
The convergence of human genomics and decentralized AI opens a new chapter for healthtech.
Questions
Frequently Asked Questions
Synthetic data can support research when it is validated against the statistical properties and biological relationships relevant to the intended use. NIOME's challenge framework is designed to make those evaluation criteria explicit rather than treating synthetic quality as a single universal claim.
NIOME aims to generate research cohorts that do not reproduce an identifiable person's genome. Project-specific privacy, governance, and regulatory requirements still need to be assessed for the intended workflow and jurisdiction.
Genomic information is persistent, uniquely sensitive, and relevant to biological relatives as well as the individual. NIOME is exploring how synthetic cohorts can reduce reliance on distributing identifiable genomic records for downstream research.
NIOME is live as Bittensor Subnet 55. Miner documentation is being prepared; join the Genomes.io Discord for current participation announcements and technical updates.
Yes. NIOME is live on Bittensor mainnet as Subnet 55. Challenge 1 is complete, and the current challenge is Challenge 2: HBB + CRISPR Editing Outcome.
Enterprise projects begin with a scoping discussion covering the research objective, cohort design, validation criteria, export format, and integration requirements. Contact the research team to discuss a project.
Get Started
Choose your path into genomic intelligence.
Discuss a Research Project
For subnet participation, stake in Subnet 55 ↗