AI Paper Atlas

Research map / Image generation / Fast diffusion sampling

HELICS: Biobank-scale Conditional Synthetic Genome Generation via Latent Flow Matching

Ahmad Abdel-Azim, Xihong Lin

NeurIPS 2026 · Image generation › Fast diffusion sampling · approach: Rectified flow straightening · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Proposes HELICS, a latent flow matching model for conditional synthetic genome generation at biobank scale, aiming to produce individual-level genotypes for data-limited settings.

Problem

Individual-level genotype data is scarce for underrepresented ancestries, rare or imbalanced disease cohorts, and restricted-access biobanks, and conditional synthetic genome generation requires learning an ultra-high-dimensional genotype distribution.

Approach

Uses latent flow matching to model ultra-high-dimensional genotypes in a compressed latent space, enabling conditional generation of synthetic genomes (details beyond the abstract not available).

Measured effects reported in the paper

What was changedWhereMeasuredEffect
train HELICS only on EUR samples instead of multi-ancestry data (ablation)HELICS · Chr15 EUR vs CCDGLD difference: 0.115 -> 0.067Improved
train HELICS only on EUR samples instead of multi-ancestry data (ablation)HELICS · Chr15 EUR vs CCDGlow-MAF R^2: 0.362 -> 0.493Improved
train HELICS only on EUR samples instead of multi-ancestry data (ablation)HELICS · Chr15 EUR vs CCDGMAF R^2: 0.992 -> 0.962Hurt
add MAF and LD penalties and condition-prediction loss to the autoencoderHELICS autoencoder · UKBcohort-level fidelityNo clear effect
add late-stage latent regularization (TV, robustness, stability) to the autoencoder (ablation)HELICS autoencoder · UKB validation (per chromosome)validation reconstruction MSEImproved
add late-stage latent regularization to the autoencoder (ablation)HELICS autoencoder · UKB validation (per chromosome)latent TV, robustness and stability penaltiesImproved
HELICS (genome-wide, unphased) versus Sim1000G on regional MAF matchingHELICS · 22 autosomesMAF R^2: 0.985 vs 0.980Hurt
HELICS versus HAPNEST on LD discrepancyHELICS · 22 autosomesLD difference: 0.118 -> 0.109Improved
condition on HDL PRS decile 1 to 10HELICS · UKB validationgenerated HDL PRS: Bin 10 > 1 fraction 0.98Improved
post-training spectral and PC-alignment refinement of multi-ancestry generator (ablation)HELICS · multi-ancestrylow-frequency MAF agreementMixed
fewer training samples for underrepresented ancestriesHELICS · ancestry-stratified UKBlatent alignment and fidelityHurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy