omicverse-single-cell-batch-integration

Run OmicVerse single-cell batch integration on preprocessed AnnData objects.

13|2|Updated Apr 5, 2026
One-click install
npx skills add https://github.com/omicverse/omicverse-skills --skill omicverse-single-cell-batch-integration-omicverse
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: omicverse-single-cell-batch-integration
Source: https://github.com/omicverse/omicverse-skills/tree/main/src/omicverse_skills/skills/single-cell-batch-integration
Command: npx skills add https://github.com/omicverse/omicverse-skills --skill omicverse-single-cell-batch-integration-omicverse

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill requires anndata, numpy, pandas, omicverse, scib_metrics, and includes scripts (resource) and references (resource) and assets (resource) components.

What problem does it solve?

This Skill turns post-preprocessing single-cell batch correction into a reusable workflow so you can integrate samples, compare correction backends, and verify results without rebuilding the analysis each time.

Core Features & Use Cases

It supports multiple integration strategies including Harmony, Combat, Scanorama, scVI, CellANOVA, and Concord for preprocessed AnnData objects with batch labels. It can produce and validate integrated embeddings, preserve backend-specific outputs such as X_pca_harmony, X_combat, X_scVI, and X_concord, and run benchmarking with batch and biology metrics. It is useful when you need to choose the right correction method for a dataset, recover a shared latent space, or compare integration quality across methods.

Quick Start

Use the single-cell-batch-integration skill to run the appropriate batch correction backend on your preprocessed AnnData with batch labels, then check the integrated embedding and benchmark the results.

Frequently Asked Questions about omicverse-single-cell-batch-integration

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I correct batch effects in single-cell data using multiple integration methods?

To correct batch effects in single-cell data, you can run workflows like Harmony, Combat, Scanorama, scVI, CellANOVA, and Concord on preprocessed AnnData objects with batch labels to produce integrated embeddings.

What is the best way to benchmark single-cell batch integration results?

Benchmarking single-cell batch integration involves evaluating integrated embeddings against biological and batch metrics using a Benchmarker, requiring available embedding keys to compare integration quality across different correction methods.

Does scVI batch correction require preserved raw counts in the AnnData object?

Yes, scVI batch correction requires preserved counts within the AnnData object to accurately model the data and generate the integrated latent space representation.

Can I use Scanorama for single-cell batch correction with continuous batch variables?

No, Scanorama requires categorical batch labels within the AnnData object to properly align and integrate single-cell samples across different batches.

What do I need to run CellANOVA batch correction on preprocessed single-cell data?

To run CellANOVA batch correction on preprocessed single-cell data, you must provide a control_dict in addition to the standard AnnData object with batch labels.

How does Combat compare to Harmony for single-cell batch integration?

Combat and Harmony both correct batch effects in single-cell data but produce distinct embeddings like X_combat and X_pca_harmony, allowing you to benchmark and choose the right correction method for your dataset.