Taking risks and getting there first: Interview with Jay Flatley, CEO, Illumina
From Issue 5 of FLG Magazine, we put our reader’s questions to Illumina CEO Jay Flatley on long reads, short reads, and whether we should still be referring to “next generation” sequencing. What would you ask Jay?
With the dominant sequencing platform, Illumina are responsible for making the march toward precision medicine possible. They have made fast, affordable and accurate sequencing a reality. Hitting the $1000 genome was an impressive milestone, but they’re still working on getting that price down even lower. With the sequencing market sitting firmly in the palm of their hand, which lands are left to conquer?
What do most sequencing laboratories have in common? Illumina. The California-based company holds an impressive share of the sequencing market, with an estimated 90% of all DNA data produced attributable to Illumina machines. So what do you do when you’ve got your market sewn up? For Illumina, the answer is simple – keep improving and open up new markets. They’re keeping busy developing the chemistry that will help introduce new sequencing applications into the clinic, and refining a business model designed to crack the consumer market. There are exciting things happening at Illumina, and a lot of questions around those plans. Who better to ask them to than the captain of the ship himself, CEO, Jay Flatley?
Alka Chaubey, Greenwood Genetic Center: Illumina’s NGS technology has revolutionised the sequencing era; what will be the next step in getting NEXT-XT(extra) GS technologies to address the limitations of the current methodologies?
JF: We’re continuing to make great strides in enhancing the fundamental technology we have in the market today. We generally refer to that as SBS chemistry ‘sequencing by synthesis’. That chemistry has tremendous additional headroom, so we are working to continue to improve it in every dimension: accuracy, costs, performance, genome coverage. We also have parallel work in our research department on alternative chemistries that may have additional benefits. None of those chemistries have ever come to market because they frankly don’t have the performance characteristics that SBS brings. We have a very high bar in the market place for the chemistry, and until we have something that would be better we wouldn’t launch a product based on an alternative. There’s a lot of work going on around the world by us, and many others, to get around the limitations of the existing chemistry. There are still parts of the genome that are hard to get to, we’d all like to sequence faster and less expensively – that’s going to take a lot of time and research dollars to get us there, but we’re going to continue to work on it.
Jonathan Bingham, Google Genomics: What do you think about companies like Genia Technologies? Are they viable technologies that are going to make an impact in the sequencing market?
JF: There’s a whole host of single-molecule sequencing technologies using semiconductors or nanopores. They all have particular advantages and disadvantages. Some have advantages in speed or read length, but might have dramatic disadvantages in accuracy. We think there will be a number of these technologies that come to market in some form. I think they’ll have applicability in specific markets where the particular advantages those approaches bring matter. In the bulk of the market, as we see it today, accuracy is incredibly important. Dropping from 99.9+% accuracy down to 95% will limit the market applicability of a technology. So, these are companies and technologies that we track, and have our own internal development programs working in these areas as well. I think it’s all very exciting that the field is moving forward.
Keith Bradnam, UC Davis: When Apple introduced the original iPod in 2001, it was an expensive luxury ($400) that went on to change an entire industry. Remarkably, by 2006 the iPod product line was Apple’s largest source of revenue. Today, you can buy a cheap iPod knock-off for less than $20 and iPods now account for <1% of Apple’s revenue. Smart phones have made dedicated music players largely irrelevant. So here’s my question…is the Illumina of 2015 like the Apple of 2006? What does Illumina do when, in five to ten years’ time, everyone will be getting his or her genomes sequenced and analyzed in an automated manner for less than $100? If the HiSeq platform is Illumina’s iPod, what’s going to be your iPhone?
JF: That’s a great question! We certainly do believe in the 5-10 year time frame that the ability to sequence a genome will be available to everyone because the economics will be there and the clinical utility will be there. That will be an enormous market opportunity. The first thing I would say is, unlike the iPod which became a commodity because the actual technology could be replicated by other companies – especially the physical interface, the headphone jack and storage inside the iPod. Sequencing is quite challenging by comparison. It requires the intersection of a dramatically larger number of technologies which all have to work together in quite a complicated and sophisticated way. Having said that, we think that our sequencers need to become easier to use, need to become faster, need to become cheaper. These are all things we’re working on. We obviously can’t layout our roadmap for people today, but there will be technologies beyond the HiSeq, and those technologies will ultimately enable people to sequence their genomes at much lower prices than $1,000. The trick for Illumina, of course, is to be the company that introduces that technology, and brings the equivalent of the iPhone that largely replaced the iPod to market and that we don’t let someone else do that before we do it.
Kristian Andersen, The Scripps Research Institute: Will we be able to get longer reads with the current Illumina platform or do we have to wait for newer technologies to mature?
JF: SBS chemistry is capable of read lengths beyond where we are today. Today our longest read products are 300-400bp, but SBS has the capability over time to get up to the range of around 1,000bp. It’s unlikely we’ll ever get to 10,000bp read lengths with SBS chemistry. However, we have one technology in the market today, and many others that we are working on internally and with external parties, to create what we call synthetic long reads. These create the ability to take our shorter reads and label the fragments in a way that allow you to uniquely reassemble them at the end of the sequence run. That approach provides most of the advantages of true long reads. For approximately 90% of long read applications, these synthetic methods will be sufficient. There will be a scramble for the very small part of the market where true long reads are important, and that will require something different to SBS.
Jean-Claude Marshall, Pfizer: How do you see the NGS field positioning itself to best enable personalized medicine considering the draft guidance that the FDA has issued around LDTs (Laboratory Developed Test)?
JF: I think there’s a tremendous amount of ambiguity about where things are going with regulation of LDTs. Our current view is that it looks like the FDA is going to punt this issue to congress, and that it’s less likely that congress will end up regulating LDTs. Sequencing is different than the classical technologies that the FDA has regulated, because it is changing so quickly. The performance and economic advantages of creating an IVD (In Vitro Diagnostic) become limited. By the time an IVD gets through the FDA with the level of time and effort required to do that, LDT versions of a similar product will have advanced so much that the IVD will no longer be competitive in the marketplace. It creates a challenge for the companies in the space, as well as for the agency in terms of how we move forward to achieve regulated products in NGS. We’re working very closely with the FDA on these issues. I think they understand what the underlying challenges are here, and why NGS is different than the more traditional technologies that change at a much slower rate. How this is all going to work out, I think, is yet to be seen. We, and many others, are putting a lot of effort in trying to understand how this regulatory environment and landscape will evolve.
David Smith, Mayo Clinic: First, Jay, congratulations on Illumina breaking the $1,000 barrier for the cost of at least generating the necessary sequence for whole genome sequencing. Of course, this does not break the $1,000 barrier when one factors in sequence assembly, interpretation and data storage. What really would make WGS (as well as RNAseq and WES) much more widely adopted by both the scientific and clinical markets would be further reducing sequencing costs. For example, by simply decreasing the width of each well on your patterned flow cells by 1/2, you could in one step increase sequence output 4-fold. IF you further increased read length on for example the HiSeq 4000 to 250-300 base pairs you could together bring WGS down to $200-250. This would truly herald a dramatic increase in overall sequencing. It would also make WGS a much better approach for sequencing that WES. Does Illumina have any plans to implement these two simple things, and if so, what are your timetables?
JF: We’re working on a broad set of improvements to our technology that would improve the sequencing output and therefore reduce the costs.. The first one is increasing the density of the content on our flow cells, which has been a very key factor in how we’ve been able to improve our technology to date. We’re continuing to work very hard in that direction. It’s not quite as simple as it may sound – just reducing feature size creates a follow-on challenge that the overall signal goes down because the amount of light that comes from each feature becomes reduced. That requires much better signal processing, and requires better overall analytical and detection capabilities to get equivalent accuracy from the system. So it’s not as easy as simply reducing the pitch between our features. With respect to the second question on read length, we are working very hard in this area. We think that the market is not really demanding that we just increase read length a little bit by a few hundred bases here and there, because it increases the overall time it takes to sequence. Every base pair that gets sequenced, requires reagents so reading longer doesn’t directly reduce the cost of sequencing on a per base basis. While we continue to work on long reads for specific applications, we think most of the market is happy in the paired-end 250-400bp range.
Elaine Mardis, McDonnell Genome Institute at Washington University: Who is the heir apparent to your leadership of Illumina? While you’ve been wildly successful in growing the business, is there anyone truly qualified to take on the task of keeping this incredible momentum should you wake up one morning and decide you don’t want to do this anymore?
JF: We’ve built an incredible team here at Illumina. I’ve worked with the board over many years to bring in the great talent we have at the senior executive level at this company. While I’m very often the spokesperson, the work that this company produces, and the great products and the market development work is largely done by others. We have a terrific team that works on strategy. Several of our senior executives at the company have a terrific vision of where this market is going. So we think we have a tremendous collective group here, who could continue to operate this company under any circumstance.
Anna Middleton, Wellcome Trust Sanger Institute: What do you see as the biggest challenge facing the genetic counseling profession as genomics is mainstreamed across health services? As Illumina are responsible to a large degree for sequencing actually being accessible to the clinic now, how aware are you of the downstream consequences of this on the health professionals involved. Whose responsibility should this be to consider?
JF: We’re fully aware of the impact sequencing has one the profession in general and genetic counseling specifically. I think the challenges around that, first and foremost, are that there aren’t enough genetic counselors. There are quite a limited number of trained professionals in this area. We’re working with some outside entities to help foster the development of this profession. We think it’s going to be critically important over the next 5-10 years. I think the ability we, and others, have to improve the analytics around sequencing is critically important. It’s not going to be possible, as sequencing gets introduced into health system, for genetic counselors to spend a day per genome doing the analysis and interpretation. So we’re working very hard on improving the annotation of the genome, improving the ability to do automated interpretation, to couple that interpretation directly with phenotype so we can very rapidly reduce the potential number of variants that a genetic counselor would have to look and analyze to make a determination of what might matter clinically. This is an area that has broad attention throughout the community. We’re working on the Genomics England contract, and it’s a key component of what Genomics England is trying to accomplish in building the ecosystem that can take samples directly out of the NHS pipeline and return a clinically interpreted genome. We’re not quite at the end of that yet, but over the next few years, I think that capability will exist and continue to be improved.
Brian Dougherty: Could you give us an honest appraisal of Illumina’s support for oncology research. Considering you have the dominant sequencing platform, do you think you’re doing enough to demonstrate a strong leadership position in this area.
JF: We are putting a significant amount of work into developing oncology panels that are going to be introduced in to the RUO (research use only) market. We thought that one of the most important things we could do in that space is to help set standards. There are today, many different variations of panel in the market place that causes tremendous confusion on the part of physicians, on the part of payers, and with the FDA in terms of being able to understand which of these panels is better than another, which should be reimbursed, how do we determine the clinical utility for each one of these products. So one of the things we’re doing is bringing a standardized set of panels into the market place. This is a result of the Actionable Genome Consortium, Rick Klausner, our Chief Medical Officer put together. That drafted a set of standards around what this set of oncology panels should look like. And Illumina is producing products that will launch into the market in the next couple of quarters that are fully compatible with those standards.
Liz Harley, Front Line Genomics: The term “Next Generation Sequencing” seems to be decreasing in popularity as it is become increasingly vague. Is there an alternate term you’d like to introduce to replace NGS so we can end the online debate?
JF: Well I don’t have one on my fingertips today! But I think that’s something that we should think about and see if there’s a new term we could come up with. NGS is pretty widely understood, but it’s no longer very accurate. It was very appropriate in 2007/2008 when we were comparing the massively parallel sequence that we do, to the traditional capillary sequencing. But we’ve moved so far beyond where we were in 2008 that it probably is time for a new name. So we’ll work on that one! FLG: Is there anything else you’d like to mention to our readers? JF: Reproductive health. We think there’s a tremendous potential for the application of sequencing in the reproductive health market, and so it’s a very important area of focus for us. FLG: How much of a focus are you putting on refining some of the noninvasive prenatal tests (NIPT) you develop and the consent process around them? JF: We have an entire business unit dedicated to reproductive and genetic health. It works across the full spectrum from carrier screening on the front end through IVF (in vitro fertilization) with PGS (preimplantation genetic screening) and PGD (pre-implantation genetic diagnosis) applications, through NIPT, to rare childhood disease. We’re working across these applications from both a reimbursement and technology perspective. We’re working on how to push this technology through to direct clinical practice and how to build the clinical evidence that shows how its use could improve outcomes and economics.
More on these topics