While everyone hunts for the Holy Grail of human genomic sequencing (and other large genome critters), I’ve noted before there’s an obvious, if less sexy, market:
…microbial genomes are cheap, fast, and you can provide epidemiological relevant information to clinical laboratories, hospital networks, and public health departments. I’m not arguing that we will or should sequence everything–and today that’s not feasible–but in two or three years, I don’t see any technical hurdles to routine microbiological surveillance in hospitals. This is something already being done, just with mid-20th century technology.
And I wrote that four years ago, before programs like GenomeTrakr (genomic food-borne microbial surveillance) really got off the ground. We’ve arrived. While Illumina sequencing is very cheap (though the price versus the cost depends greatly on your cost structure model), it’s not likely to get that much cheaper or faster–right now, the major costs are personnel, DNA preparation, and the machine itself. Illumina-sequenced genomes also have problems with contiguity: that is, it’s often hard to tell if genes are linked to each other, especially the clinical relevant stuff, like virulence and antibiotic resistance genes, which are often found on plasmids* that are hard to assemble (#NotAllResistanceGenes…).
However, PacBio is very good for bacterial genomes. Compared to Illumina, it doesn’t produce very much sequence–a problem for large genomes (like boring humans). But bacterial genomes are small, so the technology produces enough sequence per run (we’ll return to this in a moment). PacBio’s large read size means that it’s very possible to assemble bacterial chromosomes and plasmids in their entirely–or at least in a very small number of pieces. The problem, until now, is that, for a single bacterial genome, PacBio produces too much sequence, to the point where the reagent cost (‘the sequencing cost’) is very high per bacterial genome (>$1,000 per bacterial genome).
So PacBio finally released a protocol that enables multiplex sequencing–the ability to sequence multiple bacterial genomes at one time:
This document describes a procedure for multiplexing 5 Mb microbial genomes up to 12-plex and 2 Mb genomes up to 16-plex, with complete genomes assemblies (<10 contigs). The workflow is compatible for both the PacBio RSII and Sequel Systems. 10kb SMRTbell libraries are constructed for each sample through shearing and Exo VII treatment before going through the DNA Damage Repair and End-Repair steps. After End-Repair, barcoded adapters are ligated to each sample. Following ligation, samples are pooled, treated with Exo III and VII, and then put through two 0.45X AMPure® PB bead purification steps. Note that size-selection using a BluePippin™ system is not required. SMRTLink v4.0 is utilized to demultiplex and assemble the genomes after sequencing….
For this procedure, the required total mass of DNA, after pooling, is 1 – 2 μg. Therefore, the required amount of sheared DNA, per microbe, going into Exo VII treatment is 1 μg divided by the number of microbes. For example, in a 12-plex library, 1 μg ÷ 12 microbes = 83 ng of sheared DNA is needed for Exo VII treatment.
Translated into English, you can sequence 12 E. coli at once, or 16 Camplyobacter. This makes the sequencing cost per bacterium very affordable. The quality of these genomes, if they’re similar to previous PacBio bacterial genomes, is quite high, both in terms of contiguity (figuring out where the genes are relative to each other) and sequence accuracy (the latter, as best as I can tell, is still a problem for Oxford Nanopore).
This is a good move for PacBio, and a pretty interesting development for bacterial genomics, especially if you’re interested in plasmids and genome architecture.
Plasmids are small, ‘mini-chromosomes’ that often can move from bacterium to bacterium, and which can also carry antibiotic resistance genes–making them medically important (though other plasmids are equally cool). Like bacterial genomes, they are a circle of DNA (in most cases…)