SAN FRANCISCO (GenomeWeb) – Nabsys said it plans to launch its electronic genome mapping instrument in the first half of next year on a limited basis followed by full commercial launch in the second half of 2018.
The company recently published two papers on the preprint BioRxiv server, one describing the genome mapping technology and the other describing structural variant calling software.
Barrett Bready, the company's CEO, said that among its early collaborators are researchers at the US Centers for Disease Control and Prevention, who have used the technology to identify structural variants between strains of Bordetella pertussis that were indistinguishable via next-generation sequencing.
The Providence, Rhode Island-based Brown University spinout has had a rocky few years. A management shake up in 2014 saw the company shift gears from its core application of genome mapping to focus on cancer — an approach that ultimately led to the firm's closure. Six months later though, Nabsys underwent a financial restructuring and re-launched with some of its original core leadership team, including Bready and CTO John Oliver.
Now, Nabsys is once again focusing on genome mapping, as well as structural variant calling, and is gearing up to launch a commercial instrument, dubbed Nabsys HD-Mapping. The small box-sized instrument consists of detector chips with Nabsys' proprietary nanochannel detection configuration, a fluidic manifold through which sample and buffer are added, and the electronics required for signal amplification. Semiconductor chips containing consumables are loaded into the machine and are disposed of after a few uses. Analysis is performed using software on a laptop or desktop computer.
Nabsys intends to sell its instrument for around $25,000 with reagents likely costing a few hundred dollars per genome. It will have a throughput of about one human genome every couple of days and a microbial genome every few minutes. Bready anticipates that the firm would increase throughput rapidly, however, since it has figured out how to configure the nanodetectors at high density.
Over the last several years the company has switched to a nanochannel configuration from nanopores, which has helped it improve the signal-to-noise ratio and enabled better control over the DNA molecules themselves. This switch has also enabled the team to construct denser configurations of nanodetectors. With the nanopores, if they were positioned too closely together, they interfered with each other, Bready said.
As described in the BioRxiv paper, after DNA is isolated, the samples are nicked with endonucleases and then tagged with sequence-specific probes. Next, the DNA is coated with a bacterial DNA binding protein, which increases the diameter of the DNA, and as such, increases the signal-to-noise ratio during detection. The DNA molecules translocate through the nanochannel and the detector is able to distinguish the tagged sites on the molecule. The DNA molecules pass through the nanochannel at a rate of 1 megabase per second, Bready said.
The tags are electronically detected and can be compared to a reference to construct genome maps. The method resolves tags as close as 300 base pairs.
In the BioRxiv paper, the team first demonstrated the technology on a phage lambda genome, about 48.5 kilobases in size. The researchers showed that the system tagged the genome where expected and demonstrated high concordance between different nanodetectors. Next, they tested the system on an Escherichia coli genome, including both mapping to a reference as well as de novo assembly. They tested two different nicking endonucleases to create the maps and used proprietary assembly software. For each endonuclease, the assemblies covered 99.5 percent and 99.7 percent of the genome with an average coverage of 250-fold and 180-fold each.
Lex Nederbragt, a bioinformatician at the University of Oslo, who was not involved in the study said in an email that from a first look at the paper, the technology seemed to be "potentially a welcome addition to the range of technologies for long-range genomic mapping." He added that researchers will now need to figure out the "best combination of long reads and genomic maps that are optimal for our genome assembly project, as there does not yet seem to be a one-size-fits-all solution."
As previously reported, Nabsys researchers have also used its technology on reference samples from the National Institute of Standards and Technology's Genome in a Bottle consortium. And, in the BioRxiv paper, they described the technology, SV-Verify, that they used.
The SV-Verify software includes support vector machines (SVMs) for systematic and automated evaluation of putative structural variants. In the paper, they demonstrated SV-Verify on the well-characterized NA12878 reference sample from NIST's Genome in a Bottle project, as well as on another sample known as NA24385.
The SV-Verify software uses a hypothesis-based strategy to determine the likelihood that a potential deletion exists. For each putative deletion, the software identifies the chromosome, the start of the deletion, its size, and assigns it an SVM score.
The researchers first used the NA12878 sample to train their SVMs for deletion calling. They used deletions in the sample that were larger than 300 base pairs and that had been called by at least four technologies. After training the algorithms, the researchers tested it on a different genome, the NA24385. As expected, they found that the technology had a higher rate of confirmation for deletions that had also been called by more than one technology.
Bready said that the same method can be used to call other types of structural variants, such as insertions, inversions, and translocations. And ultimately, at launch, he said that the SV-Verify will have a sensitivity of more than 92 percent and a specificity of 97 percent for structural variants greater than 1,000 base pairs.
Bready said that from talking with early collaborators, researchers will use the instrument for two main purposes at launch: in combination with short-read sequencing technology to call larger structural variants and construct better assemblies; and also on its own to perform structural variant genome-wide association studies. He said that the Nabsys instrument would enable low-cost structural variant detection much the same way that microarrays enabled low-cost genome-wide detection of SNPs.
"A lot of missing genetic information is structural, and so researchers want to do the same types of studies they did with microarrays," only looking at structural variants, Bready said. The Nabsys instrument "will be accurate and cheap enough to do it at scale."
In its collaboration with the CDC, Nabsys researchers generated de novo assembled genome maps for 12 epidemic strains of B. pertussis. The strains "vary in pathogenicity and in response to vaccines," Bready said. But, short-read sequencing of the strains that respond and don't respond to vaccines did not identify any differences. With the Nabsys instrument though, "you see clearly the structural variant differences between them," he said.
When it launches, the company will have to compete with established players like BioNano Genomics, which markets the optical mapping systems Irys and Saphyr. In addition, the company would likely also compete with both Pacific Biosciences and Oxford Nanopore Technologies. Researchers have increasingly been using PacBio's Sequel and Oxford's MinIon instruments for structural variant detection and de novo assembly. Nonetheless, Bready said that he thinks there is a lot of room for improvement in the field and that Nabsys will be able to compete on both cost and performance.