Single-cell multi-modal GAN reveals spatial patterns in single-cell data from triple-negative breast cancer

• Published in: Patterns (New York, N.Y.) N.Y.), 2022-09, Vol.3 (9), p.100577-100577, Article 100577
• Main Authors: Amodio, Matthew, Youlten, Scott E., Venkat, Aarthi, San Juan, Beatriz P., Chaffer, Christine L., Krishnaswamy, Smita
• Format: Article
• Language: English
• Subjects: Deep Learning; GANs; generative adversarial networks; manifold learning; scRNAseq
• ISSN: 2666-3899; 2666-3899
• DOI: 10.1016/j.patter.2022.100577

Exciting advances in technologies to measure biological systems are currently at the forefront of research. The ability to gather data along an increasing number of omic dimensions has created a need for tools to analyze all of this information together, rather than siloing each technology into separate analysis pipelines. To advance this goal, we introduce a framework called the single-cell multi-modal generative adversarial network (scMMGAN) that integrates data from multiple modalities into a unified representation in the ambient data space for downstream analysis using a combination of adversarial learning and data geometry techniques. The framework’s key improvement is an additional diffusion geometry loss with a new kernel that constrains the otherwise over-parameterized GAN. We demonstrate scMMGAN’s ability to produce more meaningful alignments than alternative methods on a wide variety of data modalities and that its output can be used to draw conclusions from real-world biological experimental data. •Integrating data from multiple modalities into one analysis•Using data geometry to regularize cycle-consistent GANs•Quantifying uncertainty through noise augmentation Biological experimental data are increasingly being generated along multiple different axes, with new and more complex technologies specializing in particular measurements being developed every year. Measuring a single subject or system with multiple specialized data-collecting tools creates a natural interest in integrating the results of these individual instruments to form a single unified view. The model introduced here presents a computational technique designed for this purpose. With the single-cell multi-modal GAN (scMMGAN), there is an opportunity to measure along many different omic directions and synthesize the information from each into one larger understanding of the system under study. Combining single-cell data from different modalities is important for creating a holistic understanding of the system under study. This is demonstrated on a variety of example modalities with the scMMGAN model introduced in this paper.