EUVIMEDEUVIMED entdecken
Open Access · CC BY 4.0

Learning inherent genetic patterns and trait associations with deep generative models for discrete genotype simulation

Sihan Xie iD, Thierry Tribout iD, Didier Boichard iD, Blaise Hanczar iD, Julien Chiquet iD, Eric Barrey iD

GigaScience · 21.01.2026 · 15

DOI: 10.1093/gigascience/giag044 ↗

Kurz gefasst

Fachartikel in der Zeitschrift GigaScience, online erschienen am 14.04.2026. Verfasst von Sihan Xie und 5 weiteren Autor:innen. Beteiligte Einrichtung: GABI, INRAE, AgroParisTech, Université Paris-Saclay, Domaine de Vilvert, 78350 Jouy-en-Josas, France. Hauptthemen laut MEDLINE: Genotype, Deep Learning, Models, Genetic, Genetic Association Studies. Die Arbeit nennt 70 Literaturangaben. Der Volltext ist frei zugänglich (Open Access, CC BY 4.0). Gefördert von: INRAE.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

Deep generative models open new avenues for simulating realistic genomic data while preserving privacy and addressing data accessibility constraints. While previous studies have primarily focused on generating gene expression or haplotype data, this study explores generating genotype data in both unconditioned and phenotype-conditioned settings, which is inherently more challenging due to the discrete nature of genotype data.

Results

We developed and evaluated commonly used generative models, including Variational Autoencoders, Diffusion Models, and Generative Adversarial Networks, and proposed adaptation tailored to discrete genotype data. We conducted extensive experiments on large-scale datasets, including all chromosomes from cow and multiple chromosomes from human. Model performance was assessed using a well-established set of metrics drawn from both deep learning and quantitative genetics literature. Our results show that these models can effectively capture genetic patterns and preserve genotype–phenotype association.

Conclusions

As deep generative models are able to reproduce key characteristics of genotype data, they can serve as direct tools for genotype–phenotype simulation, while also enabling privacy-preserving data sharing. Our findings provide a comprehensive evaluation of these models and offer practical guidance for future research in genotype-phenotype simulation.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

GenotypeDeep LearningModels, GeneticGenetic Association StudiesHumansAnimalsGenerative Adversarial NetworksPhenotypeGenerative Artificial IntelligenceComputer SimulationAutoencoderCattle

Stichwörter der Autor:innen

SNPdeep generative modelsgenomicsgenotype-phenotype simulationquantitative genetics

Autor:innen und Institutionen

  • GABI, INRAE, AgroParisTech, Université Paris-Saclay, Domaine de Vilvert, 78350 Jouy-en-Josas, France.
  • INRAE, UMR MIA Paris-Saclay, AgroParisTech, Université Paris-Saclay, 22 place de l'Agronomie, 91120 Palaiseau, France.
  • IBISC, Univ Evry , Université Paris-Saclay, 40 rue du Pelvoux, 91020 Évry-Courcouronnes, France.

ORCID-Profile: Sihan Xie, Thierry Tribout, Didier Boichard, Blaise Hanczar, Julien Chiquet, Eric Barrey

Förderung

  • INRAE

Zitieren

APA

Xie, S., Tribout, T., Boichard, D., Hanczar, B., Chiquet, J., & Barrey, E. (2026). Learning inherent genetic patterns and trait associations with deep generative models for discrete genotype simulation. GigaScience, 15. https://doi.org/10.1093/gigascience/giag044

Vancouver

Xie S, Tribout T, Boichard D, Hanczar B, Chiquet J, Barrey E. Learning inherent genetic patterns and trait associations with deep generative models for discrete genotype simulation. GigaScience. 2026;15. doi:10.1093/gigascience/giag044

BibTeX
@article{Xie2026,
  author = {Xie, Sihan and Tribout, Thierry and Boichard, Didier and Hanczar, Blaise and Chiquet, Julien and Barrey, Eric},
  title = {Learning inherent genetic patterns and trait associations with deep generative models for discrete genotype simulation},
  journal = {GigaScience},
  year = {2026},
  volume = {15},
  doi = {10.1093/gigascience/giag044}
}
Kostenlos starten →