General

• Two types of enrichment → different algorithms for peak calling: broad peaks (e.g. histone modifications covering entire gene bodies) or narrow peaks (e.g. transcription factor binding)
  • Narrow peaks = easier to detect because generally stronger signal vs background

macs2 (model-based analysis of ChIP-seq)

Combines information of both sequencing tag position and orientation.

1. Remove redundancy — deal with duplicate reads; auto keeps a single read at each location and calculates the max using binomial distribution at that location
2. Modeling shift size
   • Tag density around true binding site should have bimodal enrichment pattern → use this to model shift size
   • Scans whole ChIP-seq dataset for enriched regions → given a sonication size (bandwidth) and a high-confidence fold-enrichment (mfold), MACS slides two bandwidth windows across genome to find regions with tags more than mfold enriched relative to a random tag genome distribution
   • Randomly samples 1000 of these peaks to calculate d, estimated fragment length
3. Peak detection — shifts each tag by d/2; slides window of size 2d across genome to search for peaks