By Julia Karow
Almost a year after presenting an outline of its droplet-based sequencing method publicly for the first time, GnuBio said last week that it has demonstrated proof of principle for its technology and plans to start shipping alpha systems to early-access users in June.
The Harvard University spinout also disclosed that it raised $8 million in private financing last year. It is working on a fast and scalable microfluidics-based sequencing platform priced at less than $50,000 that will allow users to analyze a fixed region in a single sample or a thousand samples at the same cost per sample.
GnuBio was co-founded in late 2009 by David Weitz, a professor of physics and applied physics at Harvard. The company has licensed droplet microfluidics and emulsion technology developed in Weitz's laboratory (IS 6/8/2010 and IS 8/31/2010).
Last summer, the company won its first grant — a one-year, $240,000 award under the National Human Genome Research Institute's "$1,000 Genome" program to develop its platform (IS 9/14/2010).
At the end of November, it closed an $8 million Series A financing round from a number of private investors who had previously invested in Affomix, an antibody technology company that several GnuBio founders were involved with. Illumina acquired Affomix for an undisclosed amount last year.
GnuBio co-founder and CEO John Boyce said that the funding should last about two years at the company's current burn rate.
The company has been using the funding to optimize its technology and build prototype instruments. Since most of the microfluidics technology had already been invented in Weitz's lab, the company mostly had to optimize it, as well as the sequencing chemistry, rather than develop it from scratch, Boyce said. For example, the company had to develop a double pico-injector, optimize temperature zones, and construct oligonucleotide libraries reliably.
Earlier this year, GnuBio moved into its own space, renting about 6,000 square feet — most of it lab space — in a building that is part of the University Park at MIT in Cambridge, Mass. It currently has 18 full-time employees.
GnuBio's sequencing assay consists of probe hybridization, polymerase extension, and probe displacement. Target DNA up to a kilobase in length that carries a fluorescent label at one end is injected into picoliter-sized emulsion droplets. Each droplet contains a hexamer from a universal oligonucleotide library, a fluorescent dye that identifies the hexamer, and a probe with a quencher that binds to the target and quenches the target label. The hexamer library contains all possible base combinations — about 4,600 in all, including base repeats.
A polymerase is then added to the drop via an online picoinjector, and if the hexamer has hybridized to the target, the polymerase will extend it and displace the quenching probe. This results in a fluorescent signal that is read out, along with the hexamer-identifying label, as the droplets flow past "low-cost optics." The process is then repeated for each possible hexamer and the sequence reconstructed.
Building the universal hexamer oligo library is "the biggest bottleneck now," according to Boyce, and will take several months to complete. It is "tricky" to produce high-quality hexanucleotides, he said, and also, some of the oligos need to have additional bases so they all have the same melting temperature and can cover long homopolymeric stretches.
[ pagebreak ]
So far, the company has mainly worked with Integrated DNA Technologies as its oligo supplier, with good success, Boyce said, but it is currently evaluating other oligo providers.
While building its universal oligo library, the company last month completed a proof-of-principle project with a scaled-down library, showing that it can sequence a 126-base PCR amplicon with high fidelity. With a GC-content of 58 percent and homopolymers up to six bases in length, the amplicon sequence was chosen because it would represent a challenge for other sequencing technologies.
The company said it was able to sequence the amplicon with 100 percent coverage and a per-base accuracy of 99.993 percent, using "raw and unfiltered" data from the prototype instrument. The coverage was 1,700-fold. The sequence run took "a few seconds," which was about 20 times slower than the commercial instrument is expected to be. Boyce pointed out that the experiment did not involve any algorithmic adjustments.
"Without any manipulation, to get that kind of data I think is pretty extraordinary," said Michael Phillips, director of the Montreal Heart Institute Pharmacogenomics Centre and a member of GnuBio's scientific advisory board. "They were able to get very high-quality sequence from one run," he said. "It's one run, it is what it is, but it does look like it could be scaled quite nicely."
He added that "if you can do this hard [sequence], it's not too much of a gap that you can do easier sequences."
Phillips said he believes the GnuBio platform could work well for clinical sequencing because it can analyze a limited number of targets in a few samples and still be cost effective. With current next-gen sequencing systems, several samples usually need to be pooled to get economies of scale. "That's not realistic for clinical work," he said. And even newer systems like the Ion Torrent are not suited for analyzing single samples," he said.
Justin Lamb, a senior scientist at the Broad Institute whose work focuses on gene expression profiling and another member of GnuBio's SAB, said the data he has seen is "very encouraging."
"The ability that they have demonstrated, at least to me, to produce high-quality sequence from a bit of DNA as typical as you can get is quite impressive," he told In Sequence.
He also noted that he has been impressed by how quickly the company has moved from "essentially having nothing other than an idea to actually having a prototype instrument."
Lamb added that he looks forward to working with GnuBio in the future on gene expression applications. He said the platform might be particularly suitable for analyzing expression of a limited number of genes in a large number of samples. "There is no obvious way to do that right now," he said. "If I do that on an Illumina [sequencer] or a Heliscope, it's very challenging to do the sample prep — pool all the samples and disambiguate them later."
Boyce said that while the company has not decided whether to publish the amplicon sequencing results, it is happy to share the data with potential collaborators. GnuBio is currently deciding on a larger sequencing project with a "much larger target" that it intends to publish. It also plans to run samples for collaborators.
Starting in June, it plans to place alpha instruments in the labs of a yet-to-be-determined number of early-access customers and to use their feedback to build a commercial instrument. The goal is to ship beta systems by the end of the year and commercial systems in the first quarter of 2012.
Last year, the company said its goal was to ship beta systems by the end of 2010, but that plan got delayed because the company decided to shift its focus to the applied and molecular diagnostics market rather than whole-genome sequencing, and because it took its time to raise outside funding. "Rather than taking money earlier…we decided to scale back and build the valuation of the company up," Boyce said.
While the company has not yet set any specifications for the commercial instrument, Boyce said that it will be a "fully integrated system" that will be able to sequence panels of genes — on the order of 50 or 100. Users will load genomic DNA, and the target selection will occur online and feed directly into the sequencing run. Several samples can be loaded at the same time; the workflow will be "an order of magnitude" faster than with other sequencing systems; and the data analysis will take place in real time.
"That really speaks to the applied market and the molecular diagnostic market," he said. At the moment, "we are not really going after the high-throughput market, sequencing whole genomes," although that might be a later development. In principle, the technology could also be used for other types of assays, such as protein assays, but GnuBio is currently focused on DNA sequencing.
The company has also abandoned previous plans to offer genotyping services and build a database of genotype and phenotype data for biomarker discovery, though it retains several patents in that area. "It's currently not part of our business plan," Boyce said.
Have topics you'd like to see covered in In Sequence? Contact the editor at jkarow [at] genomeweb [.] com.