NEW YORK (GenomeWeb) – Earlier this year, Complete Genomics announced the launch of a new $12 million high-throughput sequencing system, called Revolocity, that has the capacity to sequence up to 10,000 human genomes or 95,000 human exomes per year.
At the time, the company, which was acquired by China's BGI in 2013, said that the first two customers for the system will be Radboud University Medical Center in Nijmegen, the Netherlands, and Mater Health Services in Australia.
Last week, Complete Genomics announced a third customer for the system, the Epilepsy Society in the UK. It also put out performance data for the Revolocity, generated in house on a standard human genome sample, which it compared to data generated by a service provider on the same sample using the Illumina X Ten.
At the American Society of Human Genetics annual meeting in Baltimore last week, GenomeWeb caught up with CEO Cliff Reid to talk about what has happened since the BGI acquisition, and how the company plans to position the Revolocity system in the clinical research market. Below is an edited version of the conversation.
What has changed since Complete Genomics became part of BGI?
In terms of corporate governance, not very much. We're still in Mountain View, California, we're still a Delaware corporation, it's just our shareholders who changed. The main thing that's changed for us is really the business model. Before, we were a service provider, and what we have spent doing these last two and a half years since the acquisition is switching over from being a services company to being a products company, and that's a big change.
What we're selling, Revolocity, is sort of a packaged version of what we were running before internally. We have taken all our learning about automation and end-to-end solutions, software management, and all of those kinds of things, and productized that, and that's really what Revolocity is. That packaging of the sequencing turns out to be very attractive for clinical organizations.
The target capacity is 10,000 genomes or 95,000 exomes per year. It really is only for human genomes; we still haven't done any other species. We're still focused on human, and focused on clinical as opposed to basic research. These are systems that are being set up in environments where there are patients involved. That's part of the reason for the focus that we've had on the end-to-end integration and packaging.
If you go to the big research organizations, the big genome centers, they are pretty capable of buying all the component parts and putting them together. But if you go into a hospital network like Mater Health, that's not what they are about, they are about patient care. And they very much want to do clinical research, but they don't want to become experts in the systems of sequencing. They just want it to work. We spent a lot of time and energy on packaging, so it would work end to end.
Why should clinical organizations acquire their own sequencing system rather than outsource sequencing to a service provider, the business model Complete Genomics previously promoted?
It turns out there is a big difference between the clinical market and the basic research market. In the clinical research market, patients are very close to the research process, and most countries have controls and regulations about moving clinical samples out of the country.
We will continue to see BGI provide sequencing services around the world. That will continue to be a viable model, especially for the smaller organizations that have no interest or desire or resources to deploy one of these big systems internally. They will be outsourcing to local service providers. But for these really big guys, they want to own and control these systems. They have both the financial resources and the sample resources to make that a very viable model for them. If you talk to those big customers, they all expect and want to have their own local sequencing capability.
How many of your first three customers have their system up and running?
None — these are all for deployment in the first half of next year. The timing of deployment is up to them, not us; they are preparing their laboratories and getting their site set up.
The Nijmegen customers have data on their samples but those samples were run on our Revolocity systems in Mountain View. That will be the same for future customers: when they set out to deploy a system, they may have samples they want to run before their system is available, and we go ahead and run them for them. We will continue to provide that capability.
How many potential customers are there for Revolocity?
We've looked around and we counted, because you can name them all. Depending exactly on how you count, it's somewhere between 200 and 300 sites around the world. The really big sites are going to need a lot more sequencing than 10,000 genomes a year, so they are going to have multiple systems.
A lot of times, it's national healthcare systems, single-payer systems, and the hospitals associated with national healthcare systems. And in the US, it's the big IDNs, the integrated delivery networks. It's not the community hospitals — this is really not targeted at individual hospitals.
Where are these potential customers located?
They are very much global. Strikingly, our first three sales are not in the US. This is not a large statistical number, but this is about single-payer systems and how different the US health insurance industry is to the rest of the world.
Sequencing genomes and exomes is going to be valuable for a really long time. You can sequence patients and store their data and use it over and over again. In any single-payer system, those patients stay with that system, really, for life. In the US, though, people change healthcare providers, they move around. So a healthcare provider in the US is much less likely to want to do sequencing and have that data available for the next 30 years if their patients leave them in 30 months.
The major systems in the US that look more like national healthcare systems, the big IDNs, the Kaisers, the VA, and places like that, they have a reasonably stable patient population. But most of the US hospital networks, even the big ones, don't. So where you have stable patient populations within an insurance umbrella, those are the organizations that have more incentive to go ahead and start now and generate this very valuable data, because they know they will be able to use it over and over again. We're seeing it outside the US prior to it happening inside the US, which is kind of what we expected.
What's the performance of the Revolocity like?
We didn't have any data initially but we have now run some standard test samples from Coriell and just released the data. It is better data than we used to produce — the coverage is a little better, the false positive rate is a little lower, the Ti/Tv ratio is a little lower. If you look at all the major metrics around quality of data, this data is the best we have ever produced.
We also sent those test samples out to [Illumina] X Ten commercial service providers and compared the data. It shows our performance on SNPs is a little bit better, and the performance on more complex variants is about the same.
Has the fact that your read lengths are shorter than those of competitors been an issue?
That's the kind of thing researcher care about. Not because they want to run exomes and genomes but because they want to run other organisms than humans, where it really matters. For humans, read length really does not matter. What matters most is mate pairs. You really have to have the mate pairs because that's what enables you to jump over the Alu repeats in a human genome.
If you talk to PacBio, they will tell you the opposite — that there is so much in the human genome you cannot see with short reads.
There is de novo assembly and diploid sequencing, and we absolutely agree with that. And we have our own long fragment read technology for that. That is not shipping with the first version of Revolocity, and I say that with great regret; we had to cut off some features in order to ship it out the door. That is something we will include in future commercial releases. But that does accomplish the kinds of things that PacBio talks about around haplotyping, separately sequencing the two sets of parental chromosomes. It's an important thing to do in certain arenas, but it's still not mainstream in the clinical research environment. I think it's a very important future capability that only very few researchers around the world really care about today.
Can you briefly talk about the BGISEQ-500 desktop sequencer that BGI plans to launch later this month?
It's going to be announced at a BGI conference called ICG (International Conference on Genomics) this month. But it's really a China-only release this time around; it will not be sold outside of China. Part of the reason is that they are very involved with the China regulatory authorities. Those sequencers are being built in China. It's our core technology but this is made in China for China.
Has much of the technology development moved to China?
The technology is still all developed in Mountain View. We are the core NGS R&D shop for CGI [Complete Genomics] and BGI. But BGI does their own engineering of the small instrument in China, because they have their own supply chain, which is inexpensive compared to our supply chains here in the US.