Single-cell RNA sequencing has become essential for understanding cellular heterogeneity in disease and development, but analyzing atlas-scale datasets requires substantial computational infrastructure. Deep learning approaches like scvi-tools offer powerful solutions for batch correction, integration, and downstream analysis, yet implementing these methods at scale often means wrestling with GPU provisioning, dependency management, and reproducibility challenges. This webinar demonstrates how DNAnexus removes these barriers by providing managed GPU resources, pre-configured analysis environments, and seamless integration with public datasets from resources like CZ CELLxGENE.

We'll walk through a complete analysis workflow using scvi-tools on the DNAnexus platform, from data ingestion through model training to biological interpretation. Using a real-world example, we'll show how to leverage GPU-accelerated variational autoencoders for dimensionality reduction, batch integration, and differential expression analysis without local compute constraints or environment setup overhead.

Attendees will learn:

• How to access and prepare large-scale single-cell datasets from CELLxGENE and other public repositories directly within DNAnexus
• Practical implementation of scvi-tools workflows using managed GPU resources for model training and inference
• Strategies for reproducible analysis using notebook snapshots and version-controlled environments
• Downstream analysis approaches including clustering, visualization, and differential expression using learned embeddings
• When GPU acceleration provides meaningful advantages for single-cell analysis workflows