CHICAGO – French bioinformatics startup SeqOne wants to help laboratories in its home country accelerate testing for SARS-CoV-2 by repurposing its next-generation sequencing data-analysis software.
The company said that by using its approach alongside a newly developed high-throughput viral sequencing protocol, NGS labs will be able to process tens of thousands of tests per day.
When France became a global hot spot for the COVID-19 pandemic in March and the national government instituted a wide-ranging lockdown, Montpellier-based SeqOne wanted to become part of the solution by using massive computing infrastructure to significantly boost testing capacity while retaining accuracy.
"We basically put the whole company on R&D detail," SeqOne Chief Commercial Officer Jean-Marc Holder explained. "We had a task force. We were looking at every single academic publication out there, and we began to see that there were some things that we could do that would fit with our competencies and with the needs of the country."
The most common coronavirus tests rely on qPCR, a relatively low-volume method. However, many laboratories and companies are rushing to develop sequencing-based assays in an effort to increase testing capacity and perhaps provide additional information about the novel disease beyond a diagnostic answer.
SeqOne chose a massively parallel diagnostic sequencing assay for COVID-19 developed by a group led by Howard Salis, an associate professor of biological and chemical engineering at Penn State University. Called Dx-Seq, the Salis assay combines barcoded reverse transcription, PCR, and sequencing to test up to 19,200 patient samples in a single workflow.
While all but the most expensive qPCR instruments can only process a couple hundred COVID-19 tests per cycle, the Salis group employs multiple Illumina NextSeq NGS sequencers to achieve far higher volume per run.
It turned out that many of SeqOne's customers had NextSeq machines, while labs worldwide have been scrambling to find qPCR capacity.
In France, genetic medicine mostly resides in the 32 Centres Hospitaliers Universitaires (CHUs), a national network of teaching hospitals.
Holder said that SeqOne has users in about 60 percent of those hospitals. Most of these users are in hereditary and rare diseases, though many are involved in large-scale population screening, which has become more important during the pandemic.
"You need to manage the logistics of all these samples, and the people who are best suited to doing this are the private central labs" that already process thousands of tests per day, said Holder. That means that theoretically, labs with the equipment and automation could process close to 40,000 COVID-19 tests per day by implementing the Penn State workflow.
Salis said via email that he has had conversations with SeqOne, but has not formed a working relationship with the company.
"The bioinformatics component for Dx-Seq is conceptually straightforward and we've developed our own pipeline. If you design the barcodes well, the mapping and counting process is not difficult," Salis wrote.
Holder said that SeqOne had "numerous exchanges" not only with the Penn State team, but with others who had released relevant public-domain information on the novel coronavirus.
In developing its COVID-19 test processing software, SeqOne found and addressed bottlenecks in RNA extraction and in multiplexing by permitting large numbers of samples to be pooled for analysis while still tying each sample to an individual.
SeqOne tracks files from the sequencer to the bench, facilitating traceability, a process known in French as identitovigilance, which roughly translates as identity tracking.
"We were always thinking about how this fits into the process of a real clinical rutine laboratory. We're not a research tool. We're all about people who are delivering reports in a medical context for patients," Holder said.
"Our philosophy is not just to focus on the bioinformatics, but to think of the whole system and about how you could make a functional, lab-ready, integratable, production-grade solution," Holder said.
SeqOne raised €3 million ($3.2 million) in March 2019. With recent investments, the SeqOne Platform can process data on 5 million samples per day, Holder said.
The company now is awaiting certification of its pipeline from the French Ministry of Social Affairs and Health so that providers can be reimbursed for the service, according to Holder. Once that is in place, the company will deploy its COVID-19 technology module to three large labs in France that are looking to run at that volume, plus at least a couple more with smaller ambitions, Holder said.
Those labs — which Holder said he could not name due to nondisclosure agreements in this precertification stage — are processing SARS-CoV-2 tests on qPCR instruments now.
"Using any standard NextSeq sequencer, you'll be able to blow 20,000 analyses out in one go," he said. Those wanting smaller scale could run tests on Illumina MiSeq benchtop sequencers.
There is plenty of unused NGS capacity. "Because standard medical procedures have been scaled back, because of the COVID epidemic, a lot of these machines are just sitting empty," Holder said.
Each of the labs is getting the protocol in place and working with SeqOne to customize the technology to its particular needs and equipment. Those processes must be certified by France's National Reference Center in order to be eligible to be reimbursed by the government-run health insurance system.
The company also has been talking to the French government about assisting with contact tracing, Holder said.
From an epidemiological standpoint, SeqOne is able to centralize and geocode its data to look for trends such as emerging outbreak clusters.
The firm also may be able to run "spot checks" on, say, 1,000 people in a specific city or town every few days to watch for a potential second wave of infections, Holder said, particularly because the technology is able to deliver information on viral load, which may help with prioritizing contact tracing.
"You can make a rough assumption that higher viral loads might indicate just people who have been infected longer. It's not a completely correct assumption, but if you see lots of people with a high viral load, you could say, wait a minute, there's something going on here," Holder said.
Longer-term, SeqOne will build on this architecture to add screening functionality for future epidemics, according to Holder.
"Our business strategy will probably try to think about how we can present this as sort of a longer-term solution that can be just deployed whenever there's a requirement to identify biohazards or infections of some kind, be they viral or bacteriological," Holder said.
Building a bioinformatics business
SeqOne was incorporated in 2017, growing out of Centre Hospitalier Universitaire de Montpellier, part of the national CHU network. It was supported by two regional business incubators, the Montpellier Méditerranée Métropole Business and Innovation Centre and Languedoc-Roussillon Incubation, plus SATT AxLR, a technology transfer agency located at the University of Montpellier.
Founders Nicolas Philippe and Guillaume Buwalda have been friends since childhood. Philippe, the CEO, has a doctorate in bioinformatics, while CTO Buwalda came from the world of finance, where he worked in high-transactional-volume, cloud-based computer architecture.
Holder, who has a background in managing large-scale computing platforms, joined the project before it was incorporated as SeqOne.
The founders had a vision to adapt bioinformatics to the new reality of genomics in which volumes are growing and analyses becoming more complex.
"In the old days it was looking for SNPs, simple variants," Holder said. "Now we're looking at loss of heterogeneity, we are looking at [copy-number variation], we are looking at [tumor mutational burden]." Plus, as genomics has become more mainstream, there has been increased demand for quality assurance and certification of pipelines.
Together, Philppe and Buwalda developed the SeqOne platform IT architecture to address both scalability and security, which earned them a French patent.
SeqOne Platform relies on both private-key and public-key encryption. "The data hosted on our servers is invisible to us or unreadable to us unless a customer gives us temporary key access so that we can do support," Holder said.
The firm has adopted an "app store" model with numerous applications optimized for specific processes and pipelines. "If you're doing somatics, if you're doing healthy versus tumoral, if you're doing germline family trios, we have separate apps," Holder said.
Holder said that the company will work with research partners to develop specific apps or modify existing ones to their specific needs. "That means that there's a richer, more robust set of bioinformatic applications that they can call on," he said.
The platform runs in the cloud, though Holder described it as a "hybrid" model because some apps reside locally.
For example, the SeqOne Platform has a module called DataSync, which sits inside a customer's firewall to manage and automate data flow between laboratory information management systems, electronic health records, and SeqOne.
"The LIMS will tell us when the sequencing job is done. We initiate the upload and launch all the bioinformatics," Holder says. Biologists and pathologists do not get a message from SeqOne until there is a batch of patient data ready to interpret.
Last week, SeqOne released a module based on a newly developed methodology for the detection of mobile element insertions in germline DNA to support NGS-based clinical diagnostics.
Holder talked about the idea of the "long tail" in bioinformatics, based on the unwritten "80 percent rule" that a handful of pipelines and software modules generally catch all but about 20 percent of variations. SeqOne tries to fill in the gaps.
The company worked on a project involving midsized deletions, involving perhaps 75 percent of a read length. "They are too big to be considered to be indels and they're not big enough or grown up enough to be [single-nucleotide variants] or CNVs," he said.
Some systems are unable to align the remaining parts of the read, so those sequences get soft-clipped. SeqOne wrote algorithms that seek to make sense of these rare cases through k-mer comparisons.
"You take segments of a certain length and then you basically use algorithms to match that," Holder explained. "The result of this is we now have a very robust addition to our pipelines which handle these midsize deletions."
SeqOne has done similar with allele sequences.
The company's technology has been CE-marked to help customers certify the accuracy of their processes with national health authorities even as they iterate these processes.
Whenever SeqOne updates its platform or changes the pipelines in the platform, it runs both in-house and customer reference samples, then provides users with nonregression reports to add to their own certification files.
SeqOne still is not operating outside its home country, but the COVID-19 pandemic has accelerated plans to expand internationally as soon as possible, according to Holder. He said that there have been some conversations with organizations outside France, but no contracts as yet.