Lab 11: Genome scaffolding with Hi-C

Step 1: What is Hi-C sequencing?

How do we take assembled contigs and order/orient them into chromosomes? Hi-C is a powerful technique that captures interactions of chromatin.

DNA is balled up in a cell, packed and positioned in a somewhat predictable way based on interactions with nucleosomes that wrap DNA. There is order, to a degree.

DNA is wrapped around nucleosomes. Sometimes, DNA can form predictable loops, where distant pieces of DNA interact. Enhancers that increase gene expression from a distance are sometimes found here. TADs (topologically-associated domains) are regions of DNA that show a higher-than-expected level of chromatin interaction. That is, they form associations more often than expected compared to the genome average. Different regions of chromosomes, or different chromosomes altogether, can form different compartments in a nucleus of transcriptionally active (A) and repressed (B) regions. That is, there is a hidden layer of organization within a nucleus that Hi-C can uncover.

We can cross-link DNA that is in close proximity, creating covalent bonds that “attach” two DNA strands together. Usually, these two cross-linked regions are very closely linked on the same chromosome, with little physical distance between them on the linear length of a chromosome. Sometimes, these two cross-linked strands can be from physically distant parts of the same chromosome. Typically, individual homologous chromosomes tend to occupy distinct territories inside a nucleus, meaning that Hi-C can be used to phase DNA haplotypes in a genome assembly. In other words, since we assembled two fairly distinct haplotypes in our assembly, we should be able to extend some of those haplotype blocks using Hi-C.

We can use these short-range and long-range interactions within homologous chromosomes to scaffold our genome assembly into full-length chromosomes, and to phase out the haplotypes.

(with hifiasm, we did both of these things at the same time using the hi-c integrated assembly option)

We can count up the number of interactions between DNA regions from the Hi-C sequencing, and build a giant matrix of all those counts. We expect to find a large number of interactions between DNA that is close together on the same chromosome, which leads to an overwhelmingly high signal along the diagonal of a dot plot. But, we also want these longer-range interactions that allow us to link together contigs that are distant.

Step 2: QC our Hi-C data

Phase Genomics is a popular company that does fee-for-service (contract) work on genome assembly and metagenomic analyses. They provide kits where you can generate your own Hi-C data from your individual and do all the informatics yourself, or you can send them frozen leaf material and they will do 100% of the work (library prep, sequencing, and informatics/scaffolding).

First, we want to check to see how good our Hi-C data is. That is, are there short-range AND long-range interactions occurring, that will help us scaffold our genome? Phase Genomics provides a helpful QC pipeline and scripts that can quickly tell us the quality of our data, given only a few million reads, and a reference genome of something very closely related or a draft genome.

From Phase Genomics: “The best way to know if a Hi-C library worked is to look at how much long-range signal is in it. There are also several metrics which correlate with a suspicious library, such as a high number of PCR duplicates or a large number of reads which align to the same position in the genome (this happens when there are very short fragments in the library due for example to two restriction sites being very close together). Our QC script measures these quantities and makes a recommendation about the library based on the result.”

https://phasegenomics.github.io/2019/09/19/hic-alignment-and-qc.html

You should have bwa and samtools installed already, but if not, use Conda to install them. I have left a 1 million read subset of the Hi-C data in /scratch. Let’s run it on hap1 (/scratch/hifiasm.hic.gfa.hic.hap1.p_ctg.fasta). The R1 and R2 files start with pear.omni-c.2M.subset.*

Hint: You can also speed up bwa by multi-threading it….

Then, follow the instructions for installing Phase Genomics’ hic_qc program with Conda: https://github.com/phasegenomics/hic_qc