Generative Adversarial Networks for Omics Data: An Overview and Future Directions for Livestock Applications

Edgar Caballero Vargas; Tiago Bresolin;

HIGHLIGHT

Recent work also demonstrates that GAN‐generated data can support privacy‐preserving data sharing and the design of novel biological sequences and proteins. This literature review presents the foundational concepts of GANs, explores diverse applications across omics fields, and highlights novel opportunities for applying GAN approaches in livestock genomics.

RELATED

Related Papers     Patents    Grants    Venues    Experts    Cite    Share    Save

ABSTRACT

Generative adversarial networks (GANs) are artificial intelligence models designed to generate novel data that is indistinguishable from real data. Originally developed to generate images to complement existing datasets and create more diverse samples, GAN‐based data can enhance downstream analysis for machine learning and deep learning applications. GAN‐based approaches are also versatile and suitable for augmenting omics data, improving downstream analyses such as genotype prediction, biomarker detection, and protein de novo design. In addition, GANs can preserve and reconstruct genomic regions under selective pressure, increase sample sizes from underrepresented groups, and mitigate class imbalance effects in genomic analyses, transcriptomics, and single‐cell omics. Recent work also demonstrates that GAN‐generated data can support privacy‐preserving data sharing and the design of novel biological sequences and proteins. This literature review presents the foundational concepts of GANs, explores diverse applications across omics fields, and highlights novel opportunities for applying GAN approaches in livestock genomics.

 Paraphrase     Questionfy     Wikify     Simplify     Summarize