Yeast Oxidative Stress Analysis

Automated RNA-Seq workflow for dissecting the oxidative stress response of Saccharomyces cerevisiae challenged with hydrogen peroxide (PRJNA184040).

Repository Layout

yeast_stress/
├── docs/                     # Biological interpretation
│   └── reproducibility.md    # Step-by-step reproduction guide
├── metadata/                 # Sample annotations used by the workflow
├── results/                  # Example outputs (counts + QC artifacts)
├── src/                      # Nextflow pipeline + helper scripts
├── scripts/                  # Data + reference preparation helpers
├── Makefile                  # Automation entry points
├── Dockerfile                # R/edgeR container used for DE analysis
├── docker-compose.yml        # Stand-alone utility containers
├── nextflow.config           # Pipeline defaults
└── README.md

Reproducible Workflow

Requirements

1. Nextflow 22.10+
2. Either:
   • Conda/mamba (recommended): conda env create -f environment.yml
   • Docker (optional fallback): build the analysis image with docker build -t yeast-oxidative-stress-analysis . and run the pipeline with -profile docker.

Automated Setup (Makefile)

The included Makefile wraps every preparatory step. Run the following in order (requires the Conda environment or the listed tools on PATH: SRA Toolkit, pigz, curl, hisat2-build, and optionally gffread):

1. make download-data – downloads the raw FASTQs listed in metadata/samples.csv (set PARALLEL=<n> to fetch multiple SRR accessions at once and THREADS=<n> to tune fasterq-dump threads) and records SHA256 checksums.
2. make prepare-reference – pulls the sacCer3 genome + annotation and builds the HISAT2 index (files sacCer3.*.ht2 in data/genome/; set BUILD_THREADS=<n> to accelerate hisat2-build).
3. make docker-image – builds the yeast-oxidative-stress-analysis:latest container used in the edgeR step.
4. make run – executes the entire Nextflow workflow with the default parameters (uses the Conda profile by default; add PROFILE=docker to use containers).

See docs/reproducibility.md for a detailed checklist with the expected outputs from each step.

Input Data

1. Download the six FASTQ files from SRA project PRJNA184040 (SRR636633–SRR636638).
2. Place the files under data/ (default glob: data/*.fastq.gz).
3. Verify/adjust metadata/samples.csv, which defines the sample IDs, conditions, and expected FASTQ paths:

   sample_id,condition,fastq
   SRR636633,control,data/SRR636633.fastq.gz
   ...
   SRR636638,treated,data/SRR636638.fastq.gz

4. Prepare the HISAT2 genome index in data/genome/ (files named sacCer3.*.ht2) and the matching GTF annotation (default data/Saccharomyces_cerevisiae.R64-1-1.gtf). make prepare-reference handles both downloads and index creation.

scripts/download_fastqs.sh honors THREADS (passed to fasterq-dump) and PARALLEL (number of concurrent downloads), while scripts/prepare_reference.sh honors BUILD_THREADS for hisat2-build. Override these via environment variables to saturate faster storage/network.

All paths can be overridden at runtime via -params-file or --flag value. See nextflow.config for defaults.

Run

nextflow run src/main.nf -profile conda \
    --data_dir ./data \
    --results_dir ./results \
    --samplesheet ./metadata/samples.csv

The DSL2 pipeline executes:

flowchart LR
    A[metadata/samples.csv] --> B(FastQC raw)
    A --> C(fastp)
    C --> D(FastQC trimmed)
    C --> E(HISAT2 align)
    E --> F(Sort BAM + index)
    F --> G(featureCounts)
    G --> H(edgeR + visualization)
    H --> I[Biological interpretation]

Loading

1. FastQC on raw reads.
2. fastp single-end trimming; outputs JSON/HTML reports and trimmed FASTQs.
3. FastQC on trimmed reads for post-trim QC.
4. HISAT2 alignment to sacCer3 followed by samtools sorting + indexing.
5. featureCounts gene-level quantification (all BAMs jointly).
6. edgeR differential expression (R container or Conda env) generating CSV tables, DEG summaries, volcano plot, PCA, and heatmap ready for downstream interpretation.

Key outputs (under results/):

• qc/ – FastQC reports for raw/trimmed reads.
• trimmed/ – fastp-filtered FASTQs and reports.
• alignment/ – SAM, sorted BAM, and alignment summaries.
• counts/gene_counts.txt – matrix consumed by edgeR.
• edger/ – CSV summaries, DEG counts, volcano plot, PCA plot, heatmap, significant_genes.txt.
• edger/visualizations/ – Additional ggplot outputs (MA plot, bar chart of top DEGs) from src/visualizations.R.
• docs/figures/ – Curated figures (e.g., volcano plot, top-50 heatmap) for quick viewing.

Biological Highlights

The current run (SRR636633–SRR636638; control vs H₂O₂-treated) yielded 783 differentially expressed genes at p < 0.05 (679 up, 104 down). edgeR recovered hallmark oxidative-stress signatures:

• Detoxification & membrane protection – HSP30/YCR021C (log₂FC 7.15), DDR2/YOL052C-A (7.01), OYE3/YPL171C (6.67), SRX1/YKL086W (5.92), and GPX2/YBR244W (3.67) are among the strongest inducers, mirroring Yap1/Skn7-activated antioxidant programs.
• Redirection of energy metabolism – Multiple TCA-associated oxidoreductases (e.g., YDR453C, YDL169C) and glutathione-linked enzymes rise sharply, while mitochondrial transcripts such as COX3/Q0020 and the 15S rRNA gene Q0158 drop ~2.6 log₂-fold, indicating a temporary dampening of oxidative phosphorylation.
• RNA processing repression – Small nucleolar RNAs (LSR1, snR17b) and ribosome biogenesis factors show coordinated decreases, reflecting the shift from growth to stress survival.

See docs/biological_insights.md for a richer interpretation tied directly to the generated DEG tables and plots.

Gallery Highlights

Manual Utilities

• src/gene_expression.R – parameterized edgeR workflow (used in the pipeline).
• src/significant_genes.R – CLI helper for re-filtering DE tables with custom thresholds.
• src/convert_id.R – BioMart-based converter from gene symbols to systematic IDs.
• src/visualizations.R – generates MA plots and top-gene bar charts from edger_results.csv for quick gallery additions.
• Shell helpers (fastp.sh, hisat2.sh, etc.) remain for quick ad-hoc runs outside Nextflow.
• Conda environments live in envs/ and can be materialized with conda env create -f <env.yml> if you need to run tools manually.

References

• Dataset: PRJNA184040
• Tools: FastQC, fastp, HISAT2, samtools, featureCounts (Subread), edgeR.