RNA-seq-analysis

A reproducible, Mac-native pipeline for quantifying differential gene expression in BT-549 and MDA-MB-231 breast cancer cell lines after BRCA1 knockdown (KD) and/or TNFα stimulation.

---

✨ Key Features

Stage	Tool/Resource	Notes
QC	FastQC, MultiQC	Aggregated HTML report for all FASTQ files
UMI Extraction	UMI-tools	UMI-aware deduplication (if UMIs present)
Alignment	STAR	Fast, splice-aware alignment to GRCh38
Counting	featureCounts	Gene-level count matrix
Differential Expression	DESeq2 (R)	Customizable contrasts (e.g., KD vs CTRL)
Visualization	Matplotlib, Seaborn	PCA, volcano, and heatmap (top genes)
Annotation	GENCODE/Ensembl	Human genome reference & annotation

---

📁 Directory Structure

project/
├── analysis.py                 # Main pipeline
├── fastqc/                     # One dir per sample (FASTQ.gz)
├── refs/                       # Reference data & indices
│   ├── grch38-cdna-kallisto.idx
│   ├── star_index/             # STAR genome index (prebuilt)
│   └── gencode.v36.annotation.gtf
├── quant/                      # Per-sample quantification output
├── results/                    # Counts, DE tables
└── plots/                      # Figures (auto-created)

---

⚙️ Requirements

Category	Package/Resource
Python ≥3.8	matplotlib, pandas, numpy, seaborn, scikit-learn
R	Rscript, DESeq2 (≥1.38), optparse
CLI tools	fastqc, multiqc, kallisto, STAR, umi-tools, samtools, subread (featureCounts)

Installation (macOS/Homebrew)

brew install fastqc kallisto star umi-tools samtools subread r
pip install -U matplotlib pandas numpy seaborn scikit-learn
Rscript -e 'if (!requireNamespace("BiocManager", quietly=TRUE)) install.packages("BiocManager", repos="https://cloud.r-project.org/"); BiocManager::install("DESeq2")'
Rscript -e 'install.packages("optparse", repos="https://cloud.r-project.org/")'

---

📥 Reference Data & STAR Genome Index

You may use a prebuilt STAR genome index (recommended):

Option 1: Download a Prebuilt Index

cd refs/

# ENCODE prebuilt STAR index (GRCh38, GENCODE v36; reliable, fast)
wget https://www.encodeproject.org/files/ENCFF356LFX/@@download/ENCFF356LFX.tar.gz -O grch38_star_index.tar.gz
tar -xzvf grch38_star_index.tar.gz
mv starIndex star_index

# Download annotation
wget https://gdc.cancer.gov/files/public/file/gdc/GRCh38.d1.vd1/gencode.v36.annotation.gtf.gz
gunzip gencode.v36.annotation.gtf.gz

Option 2: Download from Google Cloud Public Datasets

pip install gsutil  # if not present
gsutil -m cp -r gs://gcp-public-data--broad-references/hg38/v0/star/<STAR_INDEX> ./star_index

Where the STAR_INDEX can be:

star_2.7.9a_primary_gencode_human_v27.tar

Option 3: Build Your Own (Advanced)

If you wish to build the STAR index yourself:

STAR --runThreadN 4 \
     --runMode genomeGenerate \
     --genomeDir star_index \
     --genomeFastaFiles GRCh38.primary_assembly.genome.fa \
     --sjdbGTFfile gencode.v36.annotation.gtf \
     --sjdbOverhang 99

---

🚀 Quick Start

# 1 — Clone/copy the project
cd project

# 2 — Prepare references as above (refs/)

# 3 — Place your FASTQ files → fastqc/SAMPLE/
#     Example: fastqc/CTRL-BT-BSA-1/CTRL-BT-BSA-1_R1.fastq.gz

# 4 — Run the pipeline
python3 analysis.py

Outputs:

• results/ → gene_counts.tsv, DE table, MultiQC report
• plots/ → pca.png, volcano_kd_vs_ctrl.png, heatmap_top50.png

---

🔄 Re-running / Partial Runs

• STAR + UMI-tools: skips quantification if output exists in quant/SAMPLE/SAMPLE_counts.txt
• FastQC/MultiQC: QC step is skipped if results/multiqc/multiqc_report.html exists
• Overwrite QC: Delete fastqc/*/qc/ to rerun QC

---

📝 Citation

If you use this pipeline, please cite:

• Bray et al., Nat. Biotech. 34, 525–527 (2016) — Kallisto
• Dobin et al., Bioinformatics 29, 15–21 (2013) — STAR
• Smith et al., Genome Res. 27, 491–499 (2017) — UMI-tools
• Liao et al., Bioinformatics 30, 923–930 (2014) — featureCounts
• Love et al., Genome Biol. 15, 550 (2014) — DESeq2

---

© 2025 Sean Posner