127: In silico generation of synthetic cancer genomes using generative AI

️ Episode 127: In silico generation of synthetic cancer genomes using generative AI

In this episode of PaperCast Base by Base, we explore OncoGAN, a generative AI pipeline designed to produce highly realistic synthetic cancer genomes. The study addresses the challenge of limited access to real cancer genomes due to privacy concerns by creating shareable, simulated datasets that preserve donor confidentiality.

Study Highlights:
The authors developed OncoGAN, a multimodel ensemble combining GANs, tabular variational autoencoders, and random sampling strategies trained on large-scale cancer genome datasets. The pipeline accurately reproduces somatic mutations, copy number alterations, and structural variants across tumor types, while maintaining tumor-specific mutational signatures and positional mutation patterns. Validation using DeepTumour demonstrated that the synthetic data closely mirrors real tumor genomes and can improve classification models when used to augment training datasets. Importantly, the simulated genomes are fully open access, overcoming the barriers of restricted patient data sharing and providing valuable resources for benchmarking and improving cancer genome analysis tools.

Conclusion:
OncoGAN represents a major advance in generating synthetic cancer genomes, enabling open, privacy-preserving data sharing while enhancing the development and benchmarking of genomic analysis tools.

Reference:
Díaz-Navarro, A., Zhang, X., Jiao, W., Wang, B., & Stein, L. (2025). In silico generation of synthetic cancer genomes using generative AI. Cell Genomics, 5, 100969. https://doi.org/10.1016/j.xgen.2025.100969

License:
This episode is based on an open-access article published under the Creative Commons Attribution 4.0 International License (CC BY 4.0) – https://creativecommons.org/licenses/by/4.0/

Support:
If you'd like to support Base by Base, you can make a one-time or monthly donation here: https://basebybase.castos.com/

Keywords: synthetic genomes, OncoGAN, cancer genomics, generative AI, data privacy