Abstract

Traditional dimensionality reduction techniques, while useful, often fail to capture the complex non-linear relationships present in high-dimensional data. Deep learning approaches, particularly Variational Autoencoders (VAEs), have emerged as powerful tools for unsupervised analysis of single-cell transcriptomic data. VAEs combine the representational power of neural networks with probabilistic modeling, enabling them to learn meaningful latent representations while accounting for the inherent uncertainty in biological data.

About This Material

This is a Hands-on Tutorial from the GTN which is usable either for individual self-study, or as a teaching material in a classroom.

Questions this will address

• How can we identify distinct cell populations in bone marrow single-cell data without prior labels?
• What cellular patterns and relationships can be discovered through unsupervised deep learning approaches?
• How does variational autoencoder (VAE) architecture help in dimensionality reduction and feature learning for single-cell data?

Learning Objectives

• Apply Flexynesis VAE architecture for unsupervised analysis of single-cell bone marrow data
• Perform dimensionality reduction and feature learning using deep learning methods
• Identify and interpret cellular clusters and patterns in high-dimensional single-cell datasets
• Evaluate the quality of unsupervised representations through visualization and clustering metrics