🧬 CRISPR Organoid Dependency Analysis

Complete computational pipeline for genome-wide CRISPR screening in patient-derived organoids

---

🎯 Overview

Comprehensive R pipeline for analyzing CRISPR dependency screens in organoid models, identifying essential genes, biomarkers, and therapeutic vulnerabilities.

Publication: A tumour-derived organoid biobank defines a functional map of gene dependencies | Nature (2026)

---

🚀 Quick Start

Installation

# Install CRAN packages
install.packages(c("data.table", "tidyverse", "scales", "lmtest", "digest"))

# Install Bioconductor packages
BiocManager::install(c("CoRe", "msigdbr", "clusterProfiler", "CRISPRcleanR", "sva"))

Run Pipeline

1. Process raw CRISPR data

source("scripts/CRISPR_pipeline.R") source(run_BAGEL2_bf.sh) source(run_BAGEL2_pr.sh)

    CRISPR_pipeline.R (R)
        ↓
    Normalized fold changes (TSV files)
        ↓
    run_BAGEL2_bf.sh (Bash)
        ↓
    BAGEL2 Bayes Factors (.bf files)
        ↓
    BAGEL2 Precision-Recall (.pr files)
        ↓
    CRISPR_pipeline.R (R) - Import BAGEL2 results
        ↓
    Binary essentiality calls (TSV files)

2. Identify core genes

source("scripts/core_fitness_genes_analysis.R")

3. Find differential dependencies

source("scripts/CRISPR_differential_dependency_analysis.R")

4. Prepare features

source("scripts/input_biomarker_analysis.R")

5. Discover biomarkers

source("scripts/CRISPR_biomarker_analysis.R")

6. Reproduce main figure of the paper

source(Main_figures.Rmd)

---

🔬 Key Methods

CRISPR Processing

• Normalization: CRISPRcleanR → Spline → ComBat
• QC: Replicate distance, ROC curves, Cohen's D
• Calling: BAGEL2 (FDR < 0.05)

Biomarker Discovery

• Model: Linear regression with likelihood ratio test
• Features: Mutations, CNVs, expression, clinical
• Classification: A (strong), B (moderate), C (significant)
• Thresholds: FDR < 0.05, |Δmean| > 0.5

---

📥 Input Data

Required data types:

• ✅ Raw CRISPR counts (sgRNA level) - Not provided for this project

• ✅ Somatic mutations (VCF-derived)

• ✅ Copy number alterations

• ✅ RNA-seq expression (TPM)

• ✅ Clinical metadata

• ✅ Control gene lists

• ** All imput data provided in:** figshare

---

📤 Output Files

Supplementary Tables

Table	Description
S4.2	Core fitness genes in organoids
S5	Binary essentiality calls
S6	Normalized LFCs (fitness scores)
S7.1-3	Differential dependencies by cancer type
S8.1-4	Biomarkers and priority scores

---

🔄 Reproducibility

Session Information

• session_info_R.txt

Key Methods:

• ADAM: Behan et al., Nature 2019
• CRISPRcleanR: Iorio et al., BMC Genomics 2018
• BAGEL2: Kim & Hart, Genome Medicine 2021

---

📧 Contact

Questions? Open an issue or email: ch24@sanger.ac.uk

---

Last Updated: November 2025 | Status: Reviewed