Fragmentomics: Analysis of Fragment End Motifs in cfDNA

Welcome to the official repository for the Fragmentomics Bachelor’s thesis project. This study explores the diagnostic potential of fragment end motifs (FEMs) and fragment size distributions in circulating cell-free DNA (cfDNA) using data from patients with cancer and no cancer. The repository includes the full analysis pipeline implemented in Bash and R, with scripts for filtering sequencing data, computing motif frequencies, and visualizing results.

---

📁 Repository Contents

File/Folder	Description
01_filter_bam.sh	Filters BAM files to retain high-quality, properly paired reads from autosomes only.
02_extract_motifs.sh	Extracts 5′ end 4-mer sequences from cfDNA fragments.
03_calculate_mds.R	Calculates Motif Diversity Score (MDS) per sample using normalized Shannon entropy.
04_build_motif_matrix.R	Constructs a motif frequency matrix across samples for downstream analysis.
05_heatmap_fem.R	Generates a heatmap to visualize global FEM patterns across all samples.
compare_mds_by_group.R	Compares MDS scores between cancer and non-cancer groups with Wilcoxon test.
compare_motifs_wilcox.R	Performs motif-wise Wilcoxon testing and ranks significant FEMs.
selectedMotifs_boxplot.R	Plots boxplots of selected motifs to visualize frequency shifts by group.
fem_output/	Folder containing result files like heatmaps, summary stats, and motif tables.

---

📖 Project Overview

Introduction

Fragmentation patterns in cfDNA reflect biological processes like apoptosis and chromatin structure, and may carry diagnostic information. This study investigates whether 5′ end 4-mer motifs and fragment size characteristics differ between cancer and control patients.

Objectives

• Identify statistically significant differences in FEMs between cancer and control samples.
• Explore global diversity of FEMs using MDS.
• Visualize motif enrichment via heatmaps and boxplots.
• Evaluate motifs for biomarker potential.

---

🧪 Methods Summary

1. BAM Filtering (01_filter_bam.sh)

   • Removes low-quality reads.
   • Restricts to autosomal chromosomes.
   • Prepares data for FEM extraction.

2. Motif Extraction (02_extract_motifs.sh)

   • Extracts 4-mer sequences from the 5′ ends of cfDNA fragments.
   • Uses reference genome via bedtools getfasta.

3. Motif Frequency Analysis

   • Build motif frequency matrix (04_build_motif_matrix.R).
   • Visualize FEMs in a clustered heatmap (05_heatmap_fem.R).
   • Test FEM-level differences (compare_motifs_wilcox.R).
   • Plot top motifs (selectedMotifs_boxplot.R).

4. Statistical Testing

   • Wilcoxon rank-sum tests to compare cancer vs. non-cancer.
   • Adjust p-values using Benjamini-Hochberg correction.

---

🧬 Results Summary

• Certain motifs (e.g., TGTA, TATT, TAAA) showed differential abundance between cancer and non-cancer groups.
• Global MDS did not significantly differentiate cancer status.
• Stratified heatmaps revealed clusters enriched in hematopoietic and metastatic cancers.

See figures and tables in the plots/ or fem_output/ directories for full visual results.

---

install.packages(c("ggplot2", "pheatmap", "readr", "readxl", "dplyr", "tibble", "stringr", "ggpubr", "RColorBrewer"))
## 🚀 Usage

Run the pipeline step by step or selectively:

```bash
# 1. Filter BAM files
bash 01_filter_bam.sh

# 2. Extract 4-mer FEMs
bash 02_extract_motifs.sh

# 3. Compute MDS
Rscript 03_calculate_mds.R

# 4. Build motif matrix
Rscript 04_build_motif_matrix.R

# 5. Visualize results
Rscript 05_heatmap_fem.R
Rscript compare_motifs_wilcox.R