Helics generates high-fidelity synthetic genomes from biobank data
HELICS: Biobank-scale Conditional Synthetic Genome Generation via Latent Flow Matching
Machine Learning
Summary
Genetic data from many people help scientists understand diseases and develop treatments. But access to large and diverse genetic datasets is limited, especially for some groups and rare diseases. The authors introduce HELICS, a method that creates realistic fake genomes resembling real people’s DNA, without needing complex processing steps. HELICS can also produce synthetic genomes tailored to specific ancestry or disease risk, helping researchers test methods and safely share data while protecting privacy.
What this means in practice
- •For biobank data managers: Create realistic synthetic genome datasets that preserve complex DNA relationships for privacy-preserving data sharing and benchmarking.
- •For bioinformatics platform developers: Generate synthetic genomes conditioned on ancestry and disease risk to augment training data for disease modeling tools.
- •For pharmaceutical data teams: Develop privacy-safe synthetic genome cohorts for drug discovery and genetic risk simulation.$Commercial implications: Enables selling synthetic genetic datasets tailored for pharmaceutical research while ensuring privacy compliance.
Authors
Ahmad Abdel-Azim, Xihong Lin
Abstract
Individual-level genotype data underpin genetic discovery, disease-risk modeling, and therapeutic development, yet many critical settings remain data-limited, including underrepresented ancestries, rare or imbalanced disease cohorts, and restricted-access biobanks. Conditional synthetic genome generation would support privacy-aware benchmarking, simulation, and data augmentation, but requires learning an ultra-high-dimensional discrete distribution over hundreds of thousands to tens of millions of correlated single-nucleotide polymorphisms (SNPs), or sites of DNA variation. We introduce HELICS: High-fidelity gEneration via Latent Interpolant Conditional Synthesis, a phasing-free latent flow matching framework for genome-wide SNP-level generation from biobank-scale genotype data. HELICS tokenizes chromosomes with convolutional autoencoders, reducing >600K SNPs to 4,096 continuous latent tokens, and trains a transformer parameterized conditional flow matching model over the concatenated multi-chromosome latent representation. Trained on over 417K genomes in the UK Biobank, HELICS generates genome-wide common-variant synthetic cohorts without haplotype phasing and achieves stronger fidelity to real genotypes than existing haplotype-based simulators, including high allele-frequency agreement in external evaluation and superior preservation of local and long-range SNP correlation structure. Conditioning on ancestry and genetic liability summaries enables targeted generation across genetic risk profiles; in perturbation analyses, by increasing the conditioned disease trait liability, HELICS induces SNP-level dosage changes in generated genomes that are highly correlated with published association weights. HELICS provides a scalable generative modeling framework for SNP-level genome data and a path toward conditional synthetic cohorts for data-limited genetic studies.