ezorita / Hi.C

Hi-C pipeline written in C

Hi.C a Hi-C pipeline written in C

Index

1. Download and compilation

2. Usage

3. Example

1. Download and compilation

Clone from this repository using git. Compile directly using make:

$ git clone http://github.com/ezorita/Hi.C.git
$ cd Hi.C
$ make

This will generate three binaries:

• re_digest: in-silico digestion of genomes using defined restriction enzymes.
• parse_contacts: reads mapped files and finds valid Hi-C contact pairs.
• merge_contacts: simplifies the output files of parse_contacts.

2. Usage

2.1. Mapping

For mapping, we use bwa with the following parameters (for hg19, Illumina 75bp PE):

$ bwa mem -P -k17 -U0 -L0,0 -T25 genome_index.fasta HiC-read1.fastq.gz HiC-read2.fastq.gz | samtools view -bS > HiC-mapped.bam

2.2. Digesting the genome

Before finding the contacts we need to precompute the fragments produced by the restriction enzyme used in Hi-C. To do so, run re_digest as follows:

$ re_digest [organism name] [RE name] [RE sequence] [cut fw] [cut rv]

All arguments are mandatory:

• organism: The name of the organism. You must create manually the organism in the db before performing the digestion. To do so, create a directory in the same path as re_digest called db. Inside db create another directory with the name of the organism (e.g. hg) and then place the genome of the organism in fasta format in a file called genome.fasta (yes, symbolic links are allowed). See the example below for more info.
• RE name: The name of the restriction enzyme, e.g. MboI, DpnII, EcoRI...
• RE sequence: The target sequence of the restriction enzyme, e.g. GATC (MboI, DpnII), GAATTC (EcoRI)...
• cut fw/cut rv: Cutting sites at the forward/reverse strand, from the first 5' nucleotide of the RE sequence. For instance, EcoRI produces a cut such that G'AATT,C (AATT overhang), so cut fw=1, cut rv=5.

2.3. Finding contacts

Usage

To find the contacts of your Hi-C experiment, run parse_contacts:

$ parse_contacts [organism] [RE name] [HiC-mapped.sam] [[mapq]] [[insert size]]

Mandatory arguments:

• organism: The organism as described during the digestion.
• RE name: The name of the restriction enzyme used in the experiment (must have been previously digested, see above).
• HiC-mapped.sam: The file containing the output of bwa mapping. Use <(samtools view HiC-mapped.bam) instead if you used .bam compression during the mapping process.

Optional arguments:

• mapq: The minimum mapping quality of the mapped fragments (default is 20).
• insert size: The maximum insert size (in bp) of the mapping technology (default 2000, Illumina).

Output

Running parse_contacts will print in the standard output all the valid contact pairs found in the mapping file using TADbit format:

Column	Read	Description
1	both	Read name
2	1	Chromosome
3	1	Mapping locus
4	1	Strand (0:forward/1:reverse)
5	1	Length mapped
6	1	Upstream RE site position
7	1	Downstream RE site position
8	2	Chromosome
9	2	Mapping locus
10	2	Strand (0:forward/1:reverse)
11	2	Length mapped
12	2	Upstream RE site position
13	2	Downstream RE site position

2.4. Compacting the contacts

Sorting the output file

In this step we will merge the contact pairs into pairs of RE fragments. However, first we need to sort the output file of parse_contacts with the following keys (GNU sort) -k2,2 -k8,8 -k6,6n -k12,12n, e.g.:

$ sort -k2,2 -k8,8 -k6,6n -k12,12n parse_contacts.out > parse_contacts_sorted.out

Usage

After sorting we can safely merge the contacts using merge_contacts. This scripts merges the contacts of the same restriction enzyme fragments and removes potential PCR duplicates.

$ merge_contacts [parse_contacts_sorted.out]

Mandatory arguments:

• parse_contacts_sorted.out: A file generated with parse_contacts and sorted with GNU sort.

Output

The output produced by merge_contacts is similar to a bed file. The first 6 columns are the restriction enzyme fragments of the two contacting sequences and the last column is the contact count:

Column	Description
1	Chromosome of fragment 1
2	First nucleotide of fragment 1
3	Last nucleotide of fragment 1
4	Chromosome of fragment 2
5	First nucleotide of fragment 2
6	Last nucleotide of fragment 2
7	Contact count between fragments 1 and 2

3. Example

Introduction

In this example we will process Hi-C data from human cells. The restriction enzyme used during the experiment is MboI. The DNA was sequenced in an Illumina platform using 75nt paired-end reads.

Files and paths

Assume that we have our precious paired-end reads in two files called hic-read1.fastq.gz and hic-read2.fastq.gz in the same folder as our compiled binaries re_digest, parse_contacts and merge_contacts. Also assume that we have the human genome reference file and its bwa index in /genomes/hg/genome.fasta.

Processing steps

1. Make a digestion reference of MboI for human genome.

# Create the digestion database.
$ mkdir -p db/hg
$ ln -s /genomes/hg/genome.fasta db/hg/genome.fasta

# Digest the human genome with MboI (GATC, 5' overhang: GATC)
$ ./re_digest hg MboI GATC 0 4

2. Map the sequencing reads using bwa:

$ bwa mem -P -k17 -U0 -L0,0 -T25 /genomes/hg/genome.fasta hic-read1.fastq.gz hic-read2.fastq.gz | samtools view -bS > hic-mapped.bam

3. Parse contacts from mapping file and sort the output:

$ ./parse_contacts hg MboI <(samtools view hic-mapped.bam) | sort -k2,2 -k8,8 -k6,6n -k12,12n > contacts_sorted.out

4. Merge contacts:

$ ./merge_contacts contacts_sorted.out > fragment_contacts.out