cancer-omics-pipeline

End-to-end Snakemake pipeline for pan-cancer multi-omic analysis.

Built and maintained by Scott A. Bowler — Ndhlovu Lab, Weill Cornell Medicine.

Processes paired-end FASTQ files through QC, alignment, expression quantification, differential expression, variant calling, MSI/TMB scoring, and microbiome profiling — all parallelized across HPC or cloud infrastructure via Snakemake.

---

Pipeline Overview

FASTQ → Trimmomatic → STAR / HISAT2 → HTSeq → DESeq2
                   ↘ Bowtie2 + GATK → MSI (msisensor-pro)
                                    → TMB (Mutect2 + Funcotator)
                   ↘ Pathoscope (microbiome profiling)

Module	Tool	Output
QC trimming	Trimmomatic 0.39	Trimmed FASTQ
RNA-seq alignment	STAR 2.7 / HISAT2 2.2	BAM
Read counting	HTSeq 2.0	Count TSV per sample
Differential expression	DESeq2	results.xlsx
DNA alignment + BQSR	Bowtie2 + GATK4	Recalibrated BAM
Microsatellite instability	msisensor-pro	MSI.Merged.xlsx
Tumor mutational burden	Mutect2 + Funcotator	MAF per sample
Microbiome profiling	Pathoscope 2.0	Combined.tsv

---

Quickstart

git clone https://github.com/sabowler/cancer-omics-pipeline.git
cd cancer-omics-pipeline

# 1. Edit configs/config.yaml — set your project_dir and reference paths
# 2. Edit configs/samples.tsv — add your sample names and FASTQ paths

# Dry run (check workflow without executing)
snakemake --configfile configs/config.yaml --cores 16 -n

# Full run with conda environments
snakemake --configfile configs/config.yaml --cores 16 --use-conda

# HPC run via SLURM
snakemake --configfile configs/config.yaml \
    --executor slurm \
    --jobs 50 \
    --use-conda

---

Configuration

Edit configs/config.yaml before running. Key sections:

study_id: "phs000980"
project_dir: "/data/projects/Cancer/phs000980"

references:
  genome_fasta: "/data/references/Human/GRCh38.fa"
  star_index:   "/data/references/hg38+HIV1+GFP/STAR"
  gtf:          "/data/references/hg38+HIV1+GFP/hg38HIV1GFP.gtf"
  # ... (see configs/config.yaml for full reference list)

modules:
  trimmomatic: true
  star:        true
  hisat2:      false   # Use STAR or HISAT2, not both
  htseq:       true
  deseq2:      true
  bowtie2:     true
  msi:         true
  tmb:         true
  pathoscope:  true

Sample Sheet

Edit configs/samples.tsv:

sample       type    fastq_r1                 fastq_r2                 condition
SAMPLE_001   tumor   SAMPLE_001_R1.fastq      SAMPLE_001_R2.fastq      tumor
SAMPLE_001N  normal  SAMPLE_001N_R1.fastq     SAMPLE_001N_R2.fastq     normal

Tumor/normal pairing is resolved automatically by sample name convention (SAMPLE_001 → SAMPLE_001N).

---

Output Structure

{project_dir}/
├── fastq/
│   ├── trimmed/       # Trimmomatic output
│   └── unpaired/
├── bam/               # Aligned BAM files (STAR / HISAT2 / Bowtie2+GATK)
├── htseq/             # Per-sample count files
├── deseq2/
│   ├── results.xlsx   # DESeq2 differential expression results
│   └── dds.RData      # DESeq2 dataset object
├── MSI/
│   └── MSI.Merged.xlsx
├── TMB/
│   ├── somatic/       # Raw Mutect2 VCFs
│   ├── filtered/      # FilterMutectCalls output
│   └── *.maf          # Funcotator MAF files
├── Pathoscope_GRCh38.CHM13v2/
│   ├── *-sam-report.tsv
│   └── Combined.tsv
└── logs/              # Per-rule logs

Requirements

• Snakemake ≥ 7.0
• conda or mamba (for --use-conda)
• All tool environments are defined in workflow/envs/ and installed automatically with --use-conda

---

Publications

Developed in support of PITCHER: Predictor of ImmunoTherapy response by Classifying Herv-k ExpRession (Provisional Patent #: D-11459) Co-Inventor: Lishomwa C. Ndhlovu - Weill Cornell Medicine

License

MIT