Brain Meta-Analysis Pipeline

A reproducible, scalable pipeline for performing both disease-specific and cross-phenotype meta-analyses of brain-related traits using bulk RNA-seq, GWAS summary statistics, and single-cell transcriptomic data.

---

Table of Contents

1. Motivation
2. Features
3. Repository Structure
4. Installation
5. Data Preparation
6. Workflow Overview
   • 1. Disease-Specific Analysis
   • 2. Mega Analysis Across Phenotypes
7. Outputs & Interpretation
8. Customization & Extensions
9. Best Practices
10. Contributing
11. License

---

Motivation

Brain disorders and traits share complex molecular underpinnings. While individual studies yield important insights, integrating multiple high-throughput datasets (RNA-seq, GWAS, and single-cell data) across several phenotypes uncovers shared pathways, cell-type–specific drivers, and potential therapeutic targets.

This pipeline:

• Automates uniform preprocessing and statistical modeling for each phenotype.
• Aggregates results for comparative meta-analysis to highlight convergent and divergent biology.
• Supports reproducible research by adopting standardized directory layouts and container-ready scripts.

---

Features

• Phenotype-Specific Workflows: Separate modules for Alzheimer’s, Parkinson’s, Schizophrenia, and other brain-related conditions.
• Cross-Phenotype Integration: Combine differential expression outputs, GWAS hits, and cell-type enrichment to map shared mechanisms.
• Modular Code: R scripts organized by function (preprocessing, DE analysis, integration) for easy customization.
• Data-Driven Defaults: Assumes common file formats but allows user-defined parameter overrides.
• Visualization Templates: Volcano plots, heatmaps, network diagrams, and dot plots for quick publication‑ready figures.

---

Repository Structure

Brain_meta_analysis/
├── data/
│   ├── bulk_rnaseq/        # Raw counts + metadata subfolders per phenotype
│   ├── gwas/               # GWAS summary stats per phenotype
│   └── sc_rnaseq/          # Single-cell expression & metadata
│
├── disease_specific/       # Per-phenotype analysis pipelines
│   ├── Alzheimer/
│   │   ├── config.yml      # Paths & parameters
│   │   ├── preprocessing.R
│   │   ├── differential_expression.R
│   │   └── results/        # DE tables, plots
│   ├── Parkinson/          # ... similar structure
│   └── ...                 # Additional phenotypes
│
├── mega_analysis/
│   ├── codes/
│   │   ├── integrate_rnaseq_gwas.R    # Correlate gene expression with GWAS
│   │   ├── sc_integration.R           # Map bulk results onto cell types
│   │   └── summarize_results.R        # Aggregate cross-phenotype summaries
│   └── results/                       # Combined tables & figures
│
├── scripts/              # Utility scripts (e.g., download data, format conversion)
├── notebooks/            # Jupyter notebooks for exploratory analyses
├── renv.lock             # R dependency snapshot (optional)
└── README.md             # This documentation

---

Installation

Ensure you have:

• R (>= 4.1.0) and Command-line Git
• R packages: tidyverse, data.table, limma, edgeR, DESeq2, SingleCellExperiment, Matrix, readr, ggplot2, etc.
• (Optional) renv for isolated environments

git clone https://github.com/isadeghi87/Brain_meta_analysis.git
cd Brain_meta_analysis
# (Optional) Initialize project library
Rscript -e "install.packages('renv'); renv::init(); renv::restore()"

---

Data Preparation

1. Bulk RNA-seq: Place raw count matrices (CSV/TSV) and sample metadata in data/bulk_rnaseq/PHENOTYPE/.
2. GWAS Summary Statistics: Store harmonized .tsv or .txt summary files in data/gwas/PHENOTYPE/.
3. Single-Cell Data: Provide processed expression objects (.rds or .h5ad) and cell metadata in data/sc_rnaseq/PHENOTYPE/.

Use consistent phenotype folder names (e.g., Alzheimer, Parkinson, Schizophrenia). Customize paths in each config.yml if needed.

---

Workflow Overview

The pipeline consists of two main phases:

1. Disease-Specific Analysis

For each phenotype:

Rscript disease_specific/PHENOTYPE/preprocessing.R      # Data filtering & normalization
Rscript disease_specific/PHENOTYPE/differential_expression.R  # Identify DE genes

• Output: normalized matrices, DE gene tables, volcano plots in disease_specific/PHENOTYPE/results/.

2. Mega Analysis Across Phenotypes

Once all phenotypes finish:

Rscript mega_analysis/codes/integrate_rnaseq_gwas.R
Rscript mega_analysis/codes/sc_integration.R
Rscript mega_analysis/codes/summarize_results.R

• Output: cross-phenotype gene overlap tables, cell-type enrichment heatmaps, summary networks in mega_analysis/results/.

---

Outputs & Interpretation

• DE Gene Tables: Fold changes, p-values, and adjusted p-values per phenotype
• Volcano & MA Plots: Quick visual check of expression shifts
• Cross-Phenotype Overlap: Venn diagrams or upset plots for shared genes
• Cell-Type Mapping: Dot plots showing phenotype-specific expression enriched in cell clusters
• Integrated GWAS Links: Correlation matrices linking expression signatures to GWAS hits

Refer to the generated HTML reports or RMarkdown outputs for interactive exploration.

---

Customization & Extensions

• Edit individual config.yml files to tweak filtering thresholds, fold-change cutoffs, or annotation sources.
• Add new phenotype modules by copying an existing directory under disease_specific/ and updating file paths.
• Extend notebooks/ for bespoke visualizations or downstream analyses.

---

Best Practices

• Keep raw data immutable; work only on copies in results/.
• Use version control branches for major modifications.
• Document any parameter changes in config.yml or as comments in the R scripts.

---

Contributing

Contributions are welcome! Please:

1. Fork the repository
2. Create a feature branch (git checkout -b feature/your-feature)
3. Commit your changes (git commit -m "Add new analysis module")
4. Push and open a Pull Request

See CODE_OF_CONDUCT.md for guidelines.

---

License

This project is licensed under the MIT License. See LICENSE for details.