scribble

name	scribble
domain	compbio
description	Multi-omics integration for single-cell data. Use for integrating multiple modalities like RNA, ATAC, and protein in single-cell analysis.
license	MIT license
metadata	{"skill-author":"Eric Yiru"}

Scribble: Multi-Omics Integration

Overview

Scribble is a tool/package for multi-omics integration in single-cell analysis. It provides methods for integrating different modalities (gene expression, chromatin accessibility, protein) to create unified representations of cells.

When to Use This Skill

This skill should be used when:

• Integrating multiple single-cell modalities
• Performing multi-omics integration
• Creating unified embeddings from heterogeneous data
• Analyzing paired multi-omics data

Quick Start

Installation

pip install scribble

# Or from GitHub
pip install git+https://github.com/username/scribble.git

Basic Usage

import scribble
import scanpy as sc
import numpy as np

# Load multi-modal data
rna = sc.read_h5ad("rna.h5ad")
atac = sc.read_h5ad("atac.h5ad")
protein = sc.read_h5ad("protein.h5ad")

# Integration
combined = scribble.integrate(
    [rna, atac, protein],
    method="cca"  # or "mnn", "wnn"
)

Integration Methods

CCA Integration

# Canonical Correlation Analysis
combined = scribble.integrate(
    [rna, atac],
    method="cca",
    n_components=50
)

WNN (Weighted Nearest Neighbors)

# Weighted nearest neighbors
combined = scribble.integrate(
    [rna, protein],
    method="wnn"
)

Best Practices

1. Preprocess each modality: Normalize and select features
2. Check for batch effects: Verify integration quality

Additional Resources

• Documentation: Check GitHub for latest information