usage-guide.md

Compartment Analysis - Usage Guide

Overview

This skill covers detecting A/B compartments from Hi-C data using eigenvector decomposition with cooltools.

Prerequisites

pip install cooler cooltools bioframe matplotlib

Quick Start

Tell your AI agent what you want to do:

• "Call compartments from my Hi-C data"
• "Compute A/B compartments"

Example Prompts

Compartment Calling

"Detect compartments from this cooler file"

"Compute the first eigenvector for compartment analysis"

Visualization

"Plot a saddle plot for compartment strength"

"Show the compartment track for chr1"

Comparison

"Compare compartments between treatment and control"

What the Agent Will Do

1. Load cooler at appropriate resolution (50-100kb)
2. Compute expected values
3. Compute eigenvector decomposition
4. Assign A/B compartments based on E1 sign
5. Optionally compute saddle plot

Tips

• Resolution - Use 50-100kb for compartment analysis
• GC phasing - Use GC content to correctly orient A/B
• E1 sign - Positive typically = A (active), negative = B
• Saddle plot - Shows compartmentalization strength