Former General Manager, DNA Sequencing
Name: John West
Position: Until Feb. 1, 2008, senior vice president and general manager for the DNA sequencing business, Illumina
Experience and Education:
— CEO, Solexa, 2004-2007 (until Solexa’s acquisition by Illumina)
— Vice president of DNA platforms, Applied Biosystems, 2001-2004
— Marketing director for microfluidics, Microcosm Technologies (now Coventor), 1999-2001
— President, Princeton Instruments, 1990-1998
— President and founder, BioAutomation, 1988-1990
— MBA in finance, Wharton School of Business, University of Pennsylvania, 1988
— MS in engineering, Massachusetts Institute of Technology, 1980
— BS in engineering, Massachusetts Institute of Technology, 1978
As CEO of Solexa and later general manager for Illumina’s DNA sequencing business, John West oversaw the development and commercialization of Solexa’s sequencing platform, now the Illumina Genome Analyzer.
Prior to that, he was responsible for Applied Biosystems’ DNA sequencing, gene expression, genotyping, PCR, and DNA synthesis platforms.
West recently left Illumina, and In Sequence spoke with him on the sidelines of the Advances in Genome Biology and Technology conference in Marco Island, Fla., this month about his view on next-generation sequencing technologies and their future.
What have you found most interesting at the conference so far?
There are a few things. One is, I am very happy to see people are beginning to use these next-generation sequencing tools for whole-genome sequencing. You saw the Illumina announcement about concluding its first human genome internally. I think that’s the kind of work that’s likely to expand a lot more. Elaine Mardis [at the Washington University Genome Sequencing Center] was talking at last night’s session about the cancer genome that they have been sequencing, and Richard Durbin [from the Wellcome Trust Sanger Institute] was talking about his plans [of sequencing 1,000 genomes] this morning. I think we are seeing more and more people beginning to take on these large whole human genome projects. When we started three to four years ago, people would talk about it, but it seemed pretty far in the future, and now, it is beginning to become a reality.
What has enabled this? Paired-end reads?
It’s a couple of things. One is, the paired-end reads are crucial. I think it is impractical to do a human genome resequencing project without paired ends. But also, just the scale-up, the number of customers who actually have a fair number of platforms, who have come up to speed on using the platforms and all the front-end sample prep and the bioinformatics and getting used to using the machines. They are now comfortable with taking on these larger projects. Jay Flatley, the Illumina CEO, said at the JP Morgan [Healthcare] conference [in January] that 200 instruments have been shipped to customers, so there is a certain scale. This same time last year, we announced that there had been 14 instruments shipped, so it’s over an order of magnitude increase in the installed base.
What else has struck you as interesting?
One of the things that is, to me, very compelling at this meeting is the talks you saw by Jay Shendure [from the University of Washington], and there are some other talks tonight and tomorrow, on technologies to use oligo arrays, or oligos synthesized on arrays, for partitioning the genome, to find just, for example, the exome, or other parts of the genome. I think this is a very powerful technology, because it gives people the ability to bring the cost per sample down dramatically on human samples. I think this is very important, particularly for people who want to use this technology for medical applications. I think they are going to be used a lot; I think it’s very enabling to the sequencing technology.
The cost of sequencing itself has come down a lot, but if you can just pick the part of the genome you want, that might only be 1 percent of the entire genome. Or, if you are focusing on a particular disease, it might be less than that. Now that brings the cost another two orders of magnitude, possibly, down.
There have been [several] papers [on this technology last fall]; four different groups that had different approaches to the problem. I think it shows lots of people had the same idea, and different technical approaches to achieving it. They are not perfected yet, but they are good enough that it looks quite promising. And prior to this meeting, there was very little discussion of that.
What data analysis tools have yet to be developed?
A year ago, there were not many datasets to work with, at least from the Illumina platform, and there were not any paired-end datasets to work with at all. Because paired ends are likely to be so important for all these applications, it’s important for the data analysis tools to take on the paired-end data type and use it to full advantage. It’s only really been possible to develop tools in the last month, effectively, as that kind of data became available on the Illumina platform. People have developed similar tools before, but they were only for single reads, and they will probably make paired-end versions of it.
Do you think it is possible to do de novo sequencing with short-read technologies?
Yes, with paired ends. As long as you have a combination of different insert sizes on the paired ends. But I think most of it can probably be done with fairly small inserts, because you can see a big rearrangement just by seeing the breakpoints and having a paired end span that. If you only have small paired ends, the main challenge is, it’s hard to deal with large repeats. But I think that you can get a lot of value, and probably cover most of the genome, with even fairly small insert paired ends.
It’s clear, all the platform companies, at least the major ones, are working on paired-end technologies that will have a variety of different insert lengths. The recent press release from Illumina talked about insert lengths that were 200 bases but also ones that were 2,000 bases. And if you can do 200 and 2,000, then you can presumably also do other kinds as well. And you saw 454 today talking about insert sizes of 16,000 bases. I would guess that probably within a fairly short period of time, at least 454, AB, and Illumina will all have complete capability in terms of paired-end sequencing. And the differences will have to do with the workflow and how easy to use it is and what the costs are.
Other people who might come later, like Pacific Biosciences, or Helicos, or someone else like that, they have to know, if they don’t bring paired ends right at the beginning, it’s almost not very competitive anymore. Now that there are at least three next-generation sequencing platforms out, from fairly large companies, if you are going to displace those, you want to be at least as good, and I think paired ends are so crucial that that’s almost an entry barrier.
Speaking of new platforms that are still in development, where do you see potential for these? Where might they find their market niche?
Pacific Biosciences is talking tomorrow afternoon, and I don’t know a lot about their technology, but my impression from what I have heard about them, they are getting sequencing results very rapidly. From sample to output in perhaps even just minutes. If you try to think about how to take this technology into the clinic, then, realistically, if you have to do an eight-day paired-end run, it’s a mismatch with the medical world. It’s OK for research, but if you actually want to use it clinically, the patient actually wants to know what the answer is, and sooner is better. From that standpoint ... there is probably an opportunity for something like it. It’s another possibility, as opposed to trying to go after the exact same customers. That’s probably the one I would keep an eye on the most.
Do you see any intrinsic advantage in single-molecule sequencing that Pacific, Helicos, and others are pursuing, not having to amplify the sample?
The amplification is very inexpensive, it’s a fairly negligible cost. People worry that it brings in a bias, but actually, the single-molecule approaches have bias, also. You have to remember that both the Illumina approach, which grows clusters, and also the 454 and AB approaches, which use the emulsion PCR, all start with single molecules. Each sequence was originally a single molecule, just amplified. So you have got all the statistical advantages of having many kind of individual reads, each of which just represents one molecule. Imaging just one molecule and not amplifying it, I think the advantages are probably modest, and it is more likely to be troubled by the fact that you have so little signal that the signal-to-noise may be poor, and error rates may not be very good. Helicos, for example, had a presentation [at the JP Morgan Healthcare conference] last month that showed that they miss bases; apparently there is not enough signal. It’s easier if you have a substitution error. Then, just getting some more coverage can make up for it, but if you have unintentional deletions, that’s a pretty bad kind of error. And that was, originally, the biggest problem 454 had. They have made more progress on it, but I think [companies] like Helicos are probably going to have a hard time with that.
What do you think sequencers will be able to do in five years?
I think that the existing platforms, and the ones that are coming soon, all have a lot of room for advancement. They all have sessions at this meeting where they are talking about the advances in the technology, and if you are looking where they are compared to one year ago, it’s a pretty steep curve. Although it’s been a big improvement so far, perhaps a factor of 100 over the capillary machines, my expectation is that there are a lot of opportunities in the technology, not just Illumina’s, but also AB’s and 454’s. I would expect that they would have probably at least one to two orders of magnitude more throughput five years from now. It’s hard to be more specific than that, but if you look at them, they all have the ability to go to more density, longer reads, better applications, faster imaging. There are a variety of engineering and biochemistry kinds of advances that are likely to improve the performance of these systems. That’s a pretty astonishing level of capability, and people in their laboratories, I don’t think they should aim at only what’s possible today, but it’s fair to extrapolate over at least the next couple of years.
So five years from now, all the existing platforms will still be around? Or can you see something that will supersede them?
There are a lot of technologies that people have proposed that can potentially supersede them, things like nanopores. I am not convinced that these are going to supersede them. I think the performance that these platforms are going to be able to get to is so high, it sets a really high bar. A new technology that comes in has to have amazing performance to displace these technologies.
It’s hard to say exactly which of the companies will be succeeding five years from now. Obviously, I am optimistic about the Illumina, partly because I have been involved in it, but I think you can be fairly sure that Illumina, AB, and 454, all those platforms are likely to advance by quite a bit during that five-year period. Usually, if you are going to displace an existing technology, you want to have at least one order of magnitude performance improvement. So if you think that the existing platforms can already improve one to two orders of magnitude, it means that if some other exotic technology was going to come in, it would probably have to have two to three orders of magnitude improvement over what we already have today. I think that’s a very high technological bar. It doesn’t mean it can’t be done, but I haven’t seen anything on the horizon that appears that it’s imminently going to deliver that.
What are you going to do next?
I have to look and figure out what to do next. I think it’s a great field, and I am talking to people both outside the genomics field but also looking at possibilities that would be inside the field. But I am really just beginning to look.