What modalities does the scvi-tools skill support?

The skill supports a wide range of modalities including scRNA-seq, scATAC-seq, CITE-seq (RNA + protein), spatial transcriptomics, single-cell methylation, and cytometry.

Is GPU acceleration supported when using this skill?

Yes, the skill includes best practices for utilizing GPU acceleration through PyTorch Lightning, which is essential for scaling to large datasets.

Can this skill help with batch correction in single-cell data?

Yes, it provides guidance on using models like scVI and scANVI to integrate datasets and correct for technical batch effects through covariate registration.

Does this skill help with differential expression analysis?

Yes, it assists in performing probabilistic differential expression analysis, which offers better uncertainty quantification compared to traditional frequentist methods.

How does scvi-tools integrate with Scanpy?

It is designed to be AnnData-centric, allowing for a seamless workflow where models are trained on AnnData objects and results are stored back for Scanpy-based visualization and clustering.

scvi-tools

Name: scvi-tools
Author: Activer007

byActiver007

データサイエンスとML

Analyzes single-cell omics data using deep generative models for transcriptomics, chromatin accessibility, and spatial analysis.

概要

The scvi-tools skill provides specialized guidance for using the scvi-tools Python framework to perform advanced probabilistic modeling in single-cell genomics. Built on PyTorch and PyTorch Lightning, it enables Claude to assist with complex tasks such as batch correction, dimensionality reduction, and differential expression across various modalities including scRNA-seq, scATAC-seq, and CITE-seq. This skill ensures best practices for data preprocessing in the AnnData ecosystem, specialized model selection (like scVI, totalVI, and DestVI), and efficient GPU-accelerated workflows.

主な機能

Spatially-resolved transcriptomics deconvolution and mapping
Principled differential expression and cell type annotation
Automated batch correction and latent space representation
0 GitHub stars
Probabilistic modeling for single-cell transcriptomics and chromatin data
Multimodal integration for paired/unpaired datasets and CITE-seq

ユースケース

Jointly analyzing RNA and protein expression data for multimodal cell characterization
Deconvolving spatial transcriptomics data to map specific cell types to tissue locations
Integrating heterogeneous single-cell datasets while removing technical batch effects

概要

主な機能

Spatially-resolved transcriptomics deconvolution and mapping
Principled differential expression and cell type annotation
Automated batch correction and latent space representation
0 GitHub stars
Probabilistic modeling for single-cell transcriptomics and chromatin data
Multimodal integration for paired/unpaired datasets and CITE-seq

ユースケース

Jointly analyzing RNA and protein expression data for multimodal cell characterization
Deconvolving spatial transcriptomics data to map specific cell types to tissue locations
Integrating heterogeneous single-cell datasets while removing technical batch effects