Does it work with Scanpy?

Absolutely, scvi-tools follows an AnnData-centric design pattern, allowing for seamless data exchange and downstream analysis with the Scanpy ecosystem.

Can it perform batch correction?

Yes, batch correction is a core capability of scvi-tools; technical variation is handled automatically when batch keys are registered during the setup_anndata phase.

Does this skill support GPU acceleration?

Yes, the skill provides guidance on using the accelerator='gpu' parameter to leverage CUDA for large-scale single-cell datasets within the scvi-tools framework.

Why does the skill emphasize using raw counts?

Probabilistic models in scvi-tools are designed to model the underlying count distribution directly, which is most statistically accurate when using unnormalized raw data.

Can it handle multimodal data like CITE-seq?

Yes, it includes support for models like totalVI and MultiVI specifically designed for joint analysis of different modalities like RNA and protein or chromatin accessibility.

scvi-tools

Name: scvi-tools
Author: pur3v4d3r

bypur3v4d3r

•

Data Science & ML

Analyzes single-cell omics data using deep generative models and probabilistic inference within the scvi-tools framework.

About

This skill equips Claude to perform advanced single-cell genomics analysis using the scvi-tools suite, a Python framework built on PyTorch and PyTorch Lightning. It provides expert guidance for implementing deep generative models such as scVI, scANVI, and totalVI to handle batch correction, dimensionality reduction, and multimodal integration. Whether working with scRNA-seq, scATAC-seq, or spatial transcriptomics, this skill ensures best practices are followed for probabilistic modeling, Bayesian inference, and seamless integration with the Scanpy ecosystem for data-driven biological discovery.

Key Features

Principled differential expression and cell type annotation workflows.
Deep generative modeling for single-cell RNA, ATAC, and multimodal data.
1 GitHub stars
Probabilistic batch correction and data integration across diverse datasets.
Automated setup and training of scvi-tools models using AnnData objects.
Advanced spatial transcriptomics analysis including deconvolution and mapping.

Use Cases

Mapping spatial transcriptomics data to single-cell reference atlases.
Performing joint analysis of protein and RNA expression from CITE-seq data.
Integrating multiple scRNA-seq datasets while correcting for batch effects.

About

Key Features

Principled differential expression and cell type annotation workflows.
Deep generative modeling for single-cell RNA, ATAC, and multimodal data.
1 GitHub stars
Probabilistic batch correction and data integration across diverse datasets.
Automated setup and training of scvi-tools models using AnnData objects.
Advanced spatial transcriptomics analysis including deconvolution and mapping.

Use Cases

Mapping spatial transcriptomics data to single-cell reference atlases.
Performing joint analysis of protein and RNA expression from CITE-seq data.
Integrating multiple scRNA-seq datasets while correcting for batch effects.