Can I use this for spatial transcriptomics?

Yes, it supports specific spatial models like DestVI and Stereoscope for deconvolution and mapping cell-environment relationships.

How does it handle batch correction?

It uses variational autoencoders to learn latent representations of data that account for technical covariates like batch, donor, or study ID while preserving biological variance.

Does it support GPU acceleration?

Absolutely. scvi-tools is built on PyTorch and PyTorch Lightning, and this skill provides guidance on utilizing GPU acceleration for training large-scale models.

What data format does this skill require?

This skill is optimized for the AnnData format and requires raw, unnormalized count data to function correctly with its probabilistic models.

Single-Cell Omics Analysis (scvi-tools)

Name: Single-Cell Omics Analysis (scvi-tools)
Author: Zehong-Wang

byZehong-Wang

0•

Data Science & ML

Analyzes single-cell omics data using deep generative models for batch correction, multimodal integration, and differential expression.

This skill provides expert guidance for the scvi-tools Python framework, enabling sophisticated probabilistic modeling of single-cell genomics data. It streamlines complex workflows for scRNA-seq, scATAC-seq, and spatial transcriptomics by implementing best practices for deep generative models, variational inference, and multimodal data integration. Whether performing batch correction, dimensionality reduction, or cell type annotation, this skill ensures robust, scalable, and statistically sound analysis within the AnnData and Scanpy ecosystem.

Key Features

01Probabilistic batch correction and cross-dataset integration

02Specialized workflows for scRNA-seq, scATAC-seq, and CITE-seq

03Statistical differential expression using variational inference

04Spatial transcriptomics deconvolution and mapping

05Deep generative modeling for single-cell genomics using PyTorch

060 GitHub stars

Use Cases

01Deconvolving spatial transcriptomics data to identify cell types in tissue locations

02Performing semi-supervised cell type annotation and transfer learning

03Integrating heterogeneous single-cell datasets across different laboratories or batches

What are Skills?·How to Install

Install with 🐟 Skill.Fish

npx skillfish add zehong-wang/kosmos scvi-tools

For use in Claude.ai and ChatGPT

Download Skill