Code Folder Overview

This folder contains three core .Rmd notebooks that perform distinct parts of the analysis on CTCF motif intervals and whole-genome bisulfite sequencing (WGBS) data.

---

1) CTCF_motif_overlap.Rmd

• Purpose: Creates the four TFBS categories (coad_only, coad_plus_others, common_all_8, all_w_o_coad) by overlapping single-cancer intervals. It also imports the WGBS data (normal and tumor samples) for further modeling.
• Key Steps:
  1. Reading merged/“cleaned” motif data for 8 cancers, merging intervals per cancer.
  2. Defining the logic to split COAD intervals into COAD-only, COAD+, common all 8, etc.
  3. Overlapping intervals with other data sources (e.g., JASPAR FIMO motifs) if needed.
  4. Modeling/Analysis: The majority of the final analyses for the paper (e.g., coverage-weighted regression, linear modeling with rank/fimo/nCpG, etc.) occur here.

If you want to replicate the main figures and results in the paper regarding TFBS categories in colon vs. other cancers, start with this .Rmd.

---

2) WGBS_pwd_comparison.Rmd

• Purpose: Focuses on a random sampling (e.g., 10,000 intervals) from each pairwise WGBS comparison, exploring average/median PWD (pairwise distance in DNA methylation).
• Key Steps:
  1. Randomly sampling intervals from the large WGBS dataset to reduce computational load.
  2. Calculating average, median PWD, coverage, etc. per sample pair (5 tumor pairs + 1 normal pair).
  3. Comparisons: Quick EDA or simple comparisons of normal vs. tumor PWD on the random subset.

This .Rmd is lighter than the main analysis—useful for quick checks of how PWD behaves without fully enumerating millions of intervals.

---

3) WGBS_Summarization.Rmd

• Purpose: A more general EDA (exploratory data analysis) on the WGBS data for five cancer pairs and one normal pair. Includes reading, summarizing coverage/beta values, and computing basic statistics.
• Key Steps:
  1. Summaries of coverage, depth, fraction of CpGs with ≥5 coverage, average/median Beta, etc.
  2. Summaries of the CTCF intervals from the “Science paper” dataset, e.g., the proportion of intervals that belong to each cancer, how many overlap FIMO, etc.
  3. Basic EDA: Checking distributions of coverage/beta across samples.

This notebook helps confirm data quality and gives background info before the more detailed analysis in CTCF_motif_overlap.Rmd.

---

Getting Started

1. Clone or download this repository.
2. Open each .Rmd in RStudio (or another R Markdown-compatible environment).
3. Run them in the following order if replicating the paper’s final results:
   1. WGBS_Summarization.Rmd to get an overview and basic EDA.
   2. CTCF_motif_overlap.Rmd to define the TFBS categories and run the primary modeling.
   3. WGBS_pwd_comparison.Rmd if you want a smaller sampling approach to check average/median PWD quickly.

---

Dependencies

• R version: 4.x or higher recommended.
• R Packages: data.table, dplyr, GenomicRanges, GenomeInfoDb, ggplot2, etc.
  (See each .Rmd's header or setup chunk for complete library calls.)

---

Contact

For questions about these notebooks or the data, please contact Yichen Guo at [email protected].