Deciphering Multi-way Interactions in the Human Genome

Overview

All code to perform analysis in the manuscript, Deciphering Multi-way Interactions in the Human Genome, can be found here. The code is organized into separate folders which correspond to each figure in the main manuscript and supplemental materials.

Availability: https://github.com/lindsly/Pore-C_Hypergraphs

Data Availability: Contact lindsly@umich.edu

Installation

Install the Pore-C_Hypergraphs code through one of the following methods. Both methods will create a directory named Pore-C_Hypergraphs which will contain all the necessary files for execution.

• Download the toolbox as a .zip folder and extract the contents into your MATLAB directory of choice.
• Clone this Git repository into your MATLAB directory of choice using MATLAB's built-in Source Control.
• Expected install time of code: <10 minutes
• Expected install time of MATLAB: ~30 minutes

After the Pore-C_Hypergraphs directory (and its sub-directories) are added, data must be downloaded to the folder that is responsible for creating each figure.

Code Organization

• Figure 2: Incidence matrix construction of a region in Chromosome 22 from fibroblasts. Contact frequency matrices were constructed by separating all multi-way contacts within this region of Chromosome 22 into their pairwise combinations. TADs are computed from the pairwise contacts. Multi-way contacts in this figure were determined in 100 kb resolution after noise reduction, originally derived from read-level multi-way contacts. (This code also applies to Supplemental Figure 1 for B lymphocytes). Expected run time: < 10 minutes.
• Figure 3: Incidence matrix construction of Chromosome 22 in fibroblasts. Frequencies of Pore-C contacts in Chromosome 22 can be shown in a bar plot according to the order of contact. Incidence matrix construction of the inter-chromosomal multi-way contacts between Chromosome 20 and Chromosome 22 in 1 Mb resolution. Within this figure, all data are from one fibroblast sequencing run and multi-way contacts were determined after noise reduction at 1 Mb or 100 kb resolution accordingly. Expected run time: < 10 minutes.
• Figure 4: Incidence matrix construction of the top 10 most common multi-way contacts per chromosome. Matrices are constructed at 25 Mb resolution for both fibroblasts and B lymphocytes. Specifically, 5 intra-chromosomal and 5 inter-chromosomal multi-way contacts are identified for each chromosome with no repeated contacts. If 5 unique intra-chromosomal multi-way contacts are not possible in a chromosome, they are supplemented with additional inter-chromosomal contacts. Multi-way contacts were determined in 25 Mb resolution after noise reduction. Expected run time: < 10 minutes.
• Figure 5: The most common 2-way, 3-way, 4-way, and 5-way inter-chromosome combinations for each chromosome are calculated for fibroblasts and B lymphocytes. Inter-chromosomal combinations are determined using 25 Mb resolution multi-way contacts after noise reduction and are normalized by chromosome length. Here we only consider unique chromosome instances (i.e. multiple loci in a single chromosome are ignored). Expected run time: < 10 minutes.
• Figure 6: A 5 kb region before and after each locus in a Pore-C read is queried for chromatin accessibility and RNA Pol II binding (ATAC-seq and ChIP-seq, respectively). Multi-way contacts between accessible loci that have at least 1 instance of Pol II binding are indicative of potential transcription clusters. Gene expression and transcription factor binding sites are integrated to determine potential coexpression and coregulation within multi-way contacts with multiple genes. Transcription factor binding sites are queried +/- 5 kb from the gene’s transcription start site. Expected run time: 30-60 minutes.
• Figure 7: Examples of potential transcription clusters are determined for fibroblasts and B lymphocytes. Multi-way contacts used for fibroblasts include all experiments. Examples were selected from the set of multi-way contacts summarized derived in the code for Figure 6. Expected run time: < 10 minutes.
• Figure S2: Calculates entropy of intra-chromosomal genomic hypergraph for fibroblast and B lymphocytes. Expected run time: < 10 minutes.
• Figure S3: Calculates hypergraph distance between two genome-wide hypergraphs derived from fibroblast and B lymphocytes. The background distribution is formed by measuring the hypergraph distances between the hypergraph derived from fibroblast and random hypergraphs. Also calculates hypergraph distances between intra-chromosomal genomic hypergraphs between fibroblasts and B lymphocytes. Expected run time: < 10 minutes.

System Requirements

We recommend the following system properties for all code to run properly:

• Windows 10
• At least 16 Gb RAM
  • This is highly dependent on the size of the data being analyzed
• A current version of MATLAB (2019 or later)

This code was written and tested on a Windows 10 machine with an Intel Core i7-8700 CPU and 32 Gb RAM using MATLAB R2019b.