MAGeCK Analysis - Usage Guide

MAGeCK Analysis - Usage Guide

Overview

MAGeCK is the standard tool for analyzing pooled CRISPR screens. It handles count normalization, identifies significantly enriched/depleted genes, and performs pathway analysis.

Prerequisites

pip install mageck
# or via conda
conda install -c bioconda mageck

Quick Start

Tell your AI agent what you want to do:

• "Run MAGeCK analysis on my CRISPR screen count data"
• "Compare treatment vs control samples to find essential genes"
• "Identify drug resistance genes from my screen"

Example Prompts

Basic Analysis

"I have count data from a CRISPR dropout screen. Run MAGeCK test to compare treatment vs control and find essential genes."

"Analyze my CRISPR screen with MAGeCK using median normalization. My samples are T1, T2 as treatment and C1, C2 as control."

Multi-Condition Analysis

"I have a time-course CRISPR screen with day 0, day 7, and day 14 samples. Use MAGeCK MLE to analyze all timepoints together."

"Run MAGeCK MLE on my screen data accounting for batch effects between screening rounds."

Interpretation

"I have MAGeCK results. Help me interpret the RRA scores and identify the top depleted genes."

"Compare my MAGeCK results between positive and negative selection. Which genes show the strongest effect?"

What the Agent Will Do

1. Load count data and verify format
2. Select appropriate MAGeCK mode (test vs MLE)
3. Run normalization and statistical analysis
4. Generate gene-level rankings
5. Identify significantly enriched/depleted genes
6. Create visualization of results

Tips

• Use mageck test for simple two-group comparisons
• Use mageck mle for complex designs with multiple conditions or covariates
• Median normalization is most robust for typical screens
• Quality thresholds: sgRNA mapping >70%, Gini <0.2, replicate correlation >0.8
• FDR < 0.1 for discovery, FDR < 0.05 for high confidence hits

Key Concepts

RRA Score

Robust Rank Aggregation score combining sgRNA-level statistics to gene level.

Beta Score (MLE)

Effect size from maximum likelihood estimation, comparable across genes.

References

• MAGeCK paper: doi:10.1186/s13059-014-0554-4
• MAGeCK-MLE: doi:10.1186/s13059-015-0843-6
• Documentation: https://sourceforge.net/p/mageck/wiki/Home/