SAIGEQTL

SAIGEQTL is an R package for scalable and accurate expression quantitative trait locus (eQTL) mapping for single-cell studies. It implements a Generalized Poisson mixed model to handle complex data structures and large-scale genomic data.

Key Features

• Complex data modeling: Handles multiple cells per individual and relatedness between individuals
• Discrete read counts: Specifically designed for count data from single-cell studies
• Scalable analysis: Efficient for large datasets (20k genes, 200 cell types, millions of cells/variants)
• Rare variant testing: Supports both single-variant and gene/region-based tests
• Cross-platform: Available via multiple installation methods

Installation

👉 See the complete installation guide with platform support.

Quick Start:

# Docker (works on all platforms)
docker run --rm wzhou88/saigeqtl:latest step1_fitNULLGLMM_qtl.R --help

Directory Structure

qtl/
├── scripts/          # Installation and build scripts
│   ├── install.R     # Smart installer
│   ├── install_binary.R
│   ├── cluster_install.sh
│   ├── build_*.sh    # Build scripts
│   └── update_version.sh
├── testing/          # Test and validation scripts
│   ├── run_regression_test.sh
│   ├── test_binary_*.sh
│   └── test_package_regression.R
├── config/           # Configuration files
│   ├── pixi-multi-r.toml
│   └── .saigeqtl_config
├── R/               # R source code
├── src/             # C++ source code
├── man/             # Documentation
├── extdata/         # Example data and scripts
└── binaries/        # Pre-built packages

Documentation

• 📖 User Guide: https://weizhou0.github.io/SAIGE-QTL-doc/
• 🐛 Issues: https://github.com/weizhou0/SAIGEQTL/issues

Citation

If you use SAIGEQTL in your research, please cite our paper: https://www.medrxiv.org/content/10.1101/2024.05.15.24307317v1

Quick Start

After installation, SAIGEQTL follows a 3-step workflow:

# Step 1: Fit null model
step1_fitNULLGLMM_qtl.R \
  --plinkFile=data/genotypes \
  --phenoFile=data/phenotypes.txt \
  --phenoCol=trait \
  --outputPrefix=results/step1

# Step 2: Association testing  
step2_tests_qtl.R \
  --vcfFile=data/variants.vcf.gz \
  --GMMATmodelFile=results/step1.rda \
  --varianceRatioFile=results/step1.varianceRatio.txt \
  --SAIGEOutputFile=results/step2.txt

# Step 3: Gene-level p-values
step3_gene_pvalue_qtl.R \
  --assocFile=results/step2.txt \
  --genePval_outputFile=results/gene_pvalues.txt