What data format does this skill require?

The scvi-tools skill primarily works with AnnData objects and requires unnormalized raw count data for optimal performance in its probabilistic models.

How does it handle batch effects?

It uses variational autoencoders and registered covariates during setup to learn latent representations that are automatically corrected for technical variations across different batches or studies.

Does scvi-tools support GPU acceleration?

Yes, the skill provides implementation patterns for utilizing PyTorch Lightning's GPU support to significantly accelerate model training on large datasets.

Can I use this for spatial transcriptomics?

Absolutely. It includes specific models like DestVI and Stereoscope for spatial deconvolution and mapping cell types to spatial locations.

scvi-tools Single-Cell Analysis

Name: scvi-tools Single-Cell Analysis
Author: sanand0

bysanand0

•

Data Science & ML

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

About

scvi-tools provides a powerful Python framework for probabilistic modeling of single-cell genomics data, leveraging PyTorch to implement deep generative models. This skill enables users to perform sophisticated analyses across multiple modalities, including scRNA-seq, scATAC-seq, and spatial transcriptomics, through a unified API that simplifies complex workflows such as dimensionality reduction, multi-modal integration, and spatial deconvolution. It is an essential companion for researchers needing robust, scalable, and statistically principled methods to handle technical variation and extract biological insights from high-dimensional single-cell datasets.

Key Features

Scalable batch correction and multi-study data integration
Probabilistic modeling for scRNA-seq and scATAC-seq
Multimodal analysis for CITE-seq and multi-omics data
Spatial transcriptomics deconvolution and mapping
4 GitHub stars
Deep learning-based differential expression analysis

Use Cases

Performing cell type annotation using semi-supervised learning models
Integrating heterogeneous single-cell datasets from different studies
Analyzing spatial relationships in transcriptomics data for tissue mapping

About

Key Features

Scalable batch correction and multi-study data integration
Probabilistic modeling for scRNA-seq and scATAC-seq
Multimodal analysis for CITE-seq and multi-omics data
Spatial transcriptomics deconvolution and mapping
4 GitHub stars
Deep learning-based differential expression analysis

Use Cases

Performing cell type annotation using semi-supervised learning models
Integrating heterogeneous single-cell datasets from different studies
Analyzing spatial relationships in transcriptomics data for tissue mapping