scvi-tools

Acerca de

scvi-tools is a comprehensive Python framework designed for probabilistic modeling of single-cell genomics data. Built on PyTorch and PyTorch Lightning, it provides researchers with robust deep generative models to perform essential tasks like dimensionality reduction, batch correction, and differential expression across diverse modalities including scRNA-seq, scATAC-seq, and spatial transcriptomics. This skill streamlines the entire analytical pipeline—from data registration and model training to latent representation extraction—ensuring seamless integration with the wider Scanpy ecosystem for advanced biological discovery.

Características Principales

• Probabilistic modeling for diverse single-cell modalities including RNA, ATAC, and CITE-seq
• 16 GitHub stars
• High-performance GPU acceleration for large-scale genomic datasets
• Deep generative models for unsupervised dimensionality reduction and batch correction
• Spatial transcriptomics analysis for cell type deconvolution and mapping
• Standardized API for model training, saving, and latent space extraction

Casos de Uso

• Integrating multiple single-cell datasets to remove batch effects while preserving biological variation
• Deconvoluting spatial transcriptomics data to map specific cell types within tissue samples
• Performing probabilistic differential expression analysis to identify cell-type-specific markers