Synthetic Genomics
Multi-ancestry datasets for African genomics research
Literature-grounded synthetic datasets for machine learning, method development, and capacity building—centered on Sub-Saharan Africa.
The Problem
Why Synthetic Genomics for SSA?
Under-representation
Large public genomic resources under-represent individuals of African ancestry, limiting the generalizability of risk models and variant interpretation.
Privacy Constraints
Real genomic data is highly sensitive. Cross-border sharing raises complex governance, consent, and regulatory issues for African participants.
Synthetic Accelerator
Our datasets encode distributions from literature—not real individuals. Teams can prototype models and pipelines without access to identifiable data.
Catalogue
Datasets
Explore our collection of synthetic multi-omics datasets for African populations.
DNA Methylation
electricsheepafrica/ssa-breast-dna-methylationCpG methylation beta values, promoter methylation of tumour suppressors, epigenetic age metrics.
Multi-omicsRNA Expression
electricsheepafrica/ssa-breast-rna-expressionSynthetic expression values for housekeeping genes, ER/HER2/basal markers, immune signatures.
LongitudinalIntegrated ML Dataset
electricsheepafrica/ssa-breast-integrated-mlGenomics, clinical variables, longitudinal visits, treatments, and outcomes for ML pipelines.
LongitudinalDisease Progression
electricsheepafrica/ssa-breast-disease-progressionMatched primary/metastatic tumour pairs, clonal evolution, resistance mechanisms.
LongitudinalBiomarker Trajectories
electricsheepafrica/ssa-breast-biomarker-trajectoriesCA15-3, CA27-29, and ctDNA levels over 0–24 months, stratified by response class.
PopulationSNP Array (SSA AIMs)
electricsheepafrica/ssa-snp-array-aimsAncestry-informative markers for thousands of individuals across SSA regions.
PopulationStructural Variation
electricsheepafrica/ssa-structural-variation-catalogDeletions, duplications, inversions, and complex SVs across SSA ancestry groups.
GermlineBRCA African Mutations
electricsheepafrica/brca-african-mutationsBRCA1/2 variants in African populations, including founder variants and penetrance tables.
Cross-cancerCervical HPV Genotyping
electricsheepafrica/cervical-hpv-genotyping-ssaHPV genotypes in synthetic cervical screening cohorts with SSA prevalence patterns.
Methodology
How We Build Datasets
All datasets follow a shared methodology: literature-grounded parameterisation, configuration-driven stochastic simulation, and rigorous validation.
We model dependencies between ancestry, tumour biology, treatment, and outcomes—while never exposing real individuals. Each dataset includes transparent validation reports with PASS/WARN/FAIL outcomes.
Parameters are anchored in peer-reviewed studies, reports, and registries. When literature is sparse, we use documented expert priors that are conservative and revisable.
Read Full MethodologyLiterature Inventory
Collect peer-reviewed studies and registry data
Configuration Design
YAML files encoding distributions and dependencies
Synthetic Generation
Stochastic sampling from configured distributions
Validation
Automated PASS/WARN/FAIL checks and reports
Publishing
CSV/Parquet with README cards on Hugging Face
Roadmap
What's Next
Current Release
v1 — Nov 2025
- Multi-omic breast cancer suite
- Longitudinal ML modules
- Population genomics (SNP, SV)
- Cross-cancer modules
- BRCA germline catalogue
0–12 Months
Near-term
- Enhanced ctDNA kinetics
- Prostate cancer (SSA)
- Liver and HCC modules
- Lung cancer datasets
- Fairness benchmarks
12–36 Months
Medium-term
- Pathology + genomics
- Proteomics stubs
- Teaching bundles
- Community review cycles
- Pan-African network
Collaborate
Work With Us
Join us in building representative genomics resources for Africa.
Method Developers
Use our datasets as benchmarks for survival models, fairness-aware ML, and batch correction methods.
African Institutions
Co-design modules reflecting local realities—population structure, care pathways, and risk factors.
Educators
Integrate datasets into bioinformatics curricula, bootcamps, and hackathons.
Ready to collaborate?
We welcome proposals, feedback, and suggestions for new validation checks.
Get in Touch