This article has been updated to distinguish between raw and quality-filtered output of the GenapSys platform.
NEW YORK – After several years of quiet development and several pre-announcements earlier this year, GenapSys last week officially launched its sequencing platform, a lunch box-sized $10,000 instrument that uses microfluidic CMOS chips. In addition, the company announced a $90 million Series C funding round.
The launch, which followed several months of testing with early-access users, may stir up some competition in the short-read sequencing market, which has for years been dominated by Illumina, with some market share held by Thermo Fisher Scientific, some outside the US by MGI, and very little by Qiagen, which recently dismantled its sequencing development efforts. GenapSys might eventually also compete with Oxford Nanopore's MinIon, which, like its own instrument, targets labs that haven't had access to their own sequencers yet.
Researchers at the Jackson Laboratory have tested an early version of the GenapSys system and have found it to generate high-quality data with a notably low mismatch error rate.
According to GenapSys CEO Hesaam Esfandyarpour, the company, which is based in Redwood City, California and has grown to about 100 employees, most of them in R&D, sees great demand for a low-cost sequencer that produces accurate data among small and mid-size labs that couldn't afford a sequencing instrument before.
Right now, these labs either don't use sequencing or outsource it, he said, meaning they have to deal with long turnaround times and sample batching on large instruments run in central service labs. GenapSys' instrument, on the other hand, "enables decentralization and distributed sequencing in a very unprecedented way," he said, adding that customers can obtain different kinds of throughput without having to change the instrument itself by using different sizes of chips.
Especially for clinical applications, a small platform with a low cost per sample would be very attractive, he said, because tests could be run immediately and would not be delayed because of the need to batch samples. "To get a good cost per sample today, you have to buy a really expensive machine, and every time you run it is also very expensive," he explained. "If you go to the lower-end machines, then the cost per sample gets less attractive. Our goal is really to enable that flexibility so that people don't have to batch samples on a very large scale."
Right now, the GenapSys sequencer comes with two types of chips, a 1-million-sensor chip (1M) that produces several hundred thousand reads and a 16-million-sensor chip (16M) that generates 12 to 15 million reads per run. Sometime next year, the company plans to release a 144-million-sensor chip (144M) that it is currently testing in house and that will likely produce on the order of 110 to 135 million reads per run.
With the first release, the average read length is about 150 base pairs and a run takes less than 24 hours. With further optimization, read length will increase over time, Esfandyarpour said, though he declined to speculate on the upper limit. Also, paired-end reads are currently not available but "will be released in future versions," he said. Even a read length of 150 bp to 200 bp covers many applications that only need to span the length of an exon, he added.
Along with its sequencer, the company released a $10,000 clonal amplification instrument, the GenapSys Sequencing Prep System, which has a run time of about 4 hours and requires only a few minutes of hands-on time.
A sequencing kit containing the 16M chip and a reagent cartridge costs $300 and an amplification kit for a single run on the sample prep system also costs $300, according to the firm's website.
The sequencer produces standard FASTQ files, and for the secondary data analysis, GenapSys recommends the use of its proprietary cloud solution, called GenHub, which also monitors the status of ongoing sequencing runs.
Among the early-access users of the new system is a group at the Jackson Laboratory. According to Chia-Lin Wei, director of genome technologies at Jax, the group received its first GenapSys system in February and an updated version, which is similar to the commercial instrument, in July. It currently has two instruments on site, allowing it to compare variability between machines.
For an initial benchmark study, results of which they presented in a poster at the American Society of Human Genetics annual meeting last month, the researchers sequenced microbial genomes with different GC content, a gene panel from the NA12878 human reference genome, and a glioblastoma cancer gene panel. They compared their results with Illumina MiSeq single-end read data.
Data from 10 successful GenapSys runs, using the 16M sensor chip, showed "minimal run-to-run variation" and yielded more than 10 million high-quality single-end reads of 140 bp that had a high mapping rate, according to the poster. Also, compared with the MiSeq data, the researchers found no difference in coverage distribution or GC bias.
They also found that the mismatch error rate was lower in the GenapSys data than in the MiSeq data, especially in the early parts of the reads, resulting in better SNP detection. Both companies' systems called almost all of 298 known SNPs in a human reference panel and identified few false positives.
Wei said they have not collected enough data yet to make firm conclusions about the indel or homopolymer error rate of the GenapSys.
Up until now, her lab has performed the library prep and clonal amplification manually, which has been cumbersome, but it expects to receive the GenapSys sample prep system by mid-December.
Going forward, she said, her lab plans to continue evaluating the GenapSys system and to work with Jackson's CLIA lab to develop clinical assays, in particular small gene panels that look for single nucleotide variants, for example for cancer applications.
For the platform to be ready for the clinic, though, "it still has to go through rigorous validation testing," Wei said, using samples with known results to calibrate the system. Having the automated clonal amplification system will be "very helpful" for that, she said.
Microbial genomes and small targeted panels would be good applications for the GenapSys system with the 16M chip, she said, and the 1M chip might be useful for method development projects.
GenapSys' platform is not the first short-read sequencer to use CMOS chips and electronic rather than optical detection. Ion Torrent, now part of Thermo Fisher Scientific, was the first company to launch a semiconductor chip-based sequencer, the PGM, followed by the Ion Proton, Ion S5 and S5 XL, Ion GeneStudio S5 systems, and, most recently, the Ion Torrent Genexus. Illumina launched its own semiconductor-based sequencer, called iSeq, early last year.
However, Esfandyarpour said the GenapSys technology is very different from these other systems. For example, he helped to invent the basic detection method underlying the Ion Torrent technology, which measures the release of protons, causing a change in pH, a transient signal that fades away over time. GenapSys, on the other hand, measures a change in electrical impedance that is caused by the incorporation of a nucleotide into DNA bound to a bead, which is a signal that doesn't fade away.
This has several advantages, he explained. For example, it eliminates crosstalk between adjacent sites, and the technology scales better because data don't need to be collected with a high frame rate. The approach also helps with homopolymer detection, he claimed, which has been somewhat of a challenge for optical sequencers that don't use reversible terminator nucleotides, like the 454 technology, as well as for the Ion Torrent sequencers. "The homopolymer detection is extremely comparable with a system that uses terminators," like Illumina optical sequencers, he said.
In terms of output, the GenapSys sequencer currently appears to sit somewhere between the Illumina iSeq and the Thermo Fisher Ion GeneStudio S5 systems. Jackson's Wei said her lab has generated a little more than 1 Gb of quality-filtered data per run, and 1.9 Gb of data not filtered for quality.
The iSeq 100, which has a $20,000 list price, currently produces up to 4 million filtered reads per 17.5-hour run, or 1.2 Gb of data, using 150-base paired-end reads, according to Illumina's website.
The three Ion GeneStudio S5 instruments can take a variety of chips, ranging in output from 2-3 million reads per run, or 0.6-1 Gb with 400 bp reads, to 100 to 130 million reads per run, or 20-25 Gb with 200 bp reads. Sequencing run times are between 2.5 and 4 hours, according to Thermo Fisher's website.
GenapSys' recent $90 million funding round, which was led by Foresite Capital but also included other investors, who Esfandyarpour declined to name, will go into building a commercial team, as well as expanding R&D operations.
The initial launch is restricted to the US and the company has already shipped its first instrument, he said. Next year, GenapSys plans to make the system available globally and is currently looking into different distribution options.
Wei, who has also worked with other early sequencing platforms, said she hopes this system, as well as other new technologies, will improve the sequencing field overall, and lower sequencing costs. "The only way this field moves forward is if the technology has competition," she said. "I'd like to see more of these technology platforms out there and offer people more options."