AllSeq: Driving down the cost of sequencing has been the biggest push for the NGS market up until now. Do you feel this is still the area that needs the most improvement, or are other factors becoming more important? [Continued from Part 1]
Dr. Smith: I think another impediment is that when somebody is getting into this area, there are so many questions they have to ask. “What are the platforms that we need? What is the infrastructure we need? How do we analyze the data?” And there really isn’t a simple solution. Most of the sequence providers sell machines. They don’t sell you a package that helps you take this stuff and make some sense out of it, and that’s a real problem. That’s what I was hoping Qiagen was going to be. Here is a simplified integrated solution that you could purchase, just one set of things that does everything. But it has a couple of problems. I love the idea but what they did is they ended up purchasing a sequencing [technology], sequencing by synthesis, that only sequences that one end as a relatively low output and as a result you have a machine that can just do really, really small gene panels. It’s too small for anything but the smallest gene panels. If you have ten, twenty genes I think it’s fine. But if you get into an area like the neuropathies, there’s over five hundred genes that are involved in various neuropathies – there’s no way you can do it on the Qiagen platform, so they just have a couple of modules that they’re currently using. The platform is lagging so far behind. Illumina has been out for over ten years so they have [released] dramatic improvements, but [the GeneReader] is just a new platform. I think maybe in time that Qiagen it could evolve into something like that, but I’m not too enthusiastic about an integrated system that’s dependent on a sequencing platform that isn’t competitive at all with Illumina.
AllSeq: Do you think clinical sequencing will be dominated by ‘point of care’ (POC) devices or by more centralized services?
Dr. Smith: Well that’s a great question. Now, it turns out that there are some technologies that are just about to come out that are going to be able to make these technologies more (flexible) – you could send things out and you could have a handheld sequencer. The Oxford Nanopore was supposed to be that type of an instrument. And indeed the nice thing about it is the small footprint. That’s part of the reason why they’re sending it off into space. The whole NASA thing is going to be doing sequencing in space, on those tiny little Oxford Nanopore cartridges. But right now most of sequencing is done even on the smaller footprint machines, on machines that are still relatively expensive, and as a result, right now most of the sequencing is probably going to be done in a more centralized fashion. If you’re a place like the Mayo Clinic, we’re going to certainly try to have the infrastructure here. If you’re a lab core you’ll do a similar thing. If you’re some of these smaller hospitals there really yet is not a good platform that you could just buy the machine and have everything you want. Qiagen is trying to do that. Qiagen is trying to have an integrated platform where you purchase two to three machines and you basically have everything from sample prep, library prep, small sequencing. And then on the back-end they’ve even purchased the analytical components. The idea is that you would have a complete integrated system and that system might over time be the type of system where places that don’t have a large infrastructure could kind of do it. But I’d say that right now and for the next couple of years, I think it’s still going to be dominated by the largest sequencing places and people sending stuff out.
AllSeq: Do you think the use of NGS will take off in the ‘direct to consumer’ space?
Dr. Smith: Oh it’s definitely going to happen, it’s going to be a regulatory nightmare, and there’s going to certainly be a lot of questions about what are the things that you can find. I visited a company in England called DNA Electronics. They are the ones that own the patents on the Ion Torrent platform. Jonathan Rotherberg licensed the patents, but DNA Electronics were the ones who actually developed the patents. They have a small little device that’s handheld that you just put your blood or your sample at the very top and it extracts it and it does the sequencing of it – that could be the sort of thing that really facilitates the idea of having this sort of testing at home. I would guess that in about ten years you’re going to have Japanese producing toilets that basically do some sequencing so you can look at your microbiome every time you go to the bathroom. That’s going to be very important for health and well-being.
AllSeq: You touched upon the differences between targeted, exome and whole genome sequencing earlier [see Part 1]. Can you expand upon that a bit?
Dr. Smith: Well the decision at the Mayo Clinic is based on reimbursement, so everything at the Mayo Clinic is small gene panels, because they’re very worried about (getting reimbursed). But the reality is for the cost point you’re actually better off to do an exome than to do a small gene panel. And my colleagues at Wash U., a lot of their clinical tests, instead of doing the small gene panels are exome-based tests. I think that’s a better way to do it because you can find more information, you can develop powerful databases because, again, we are at the very dawn of this era. I think that the decisions are not based on what is the best approach. It’s based on affordability and worries about reimbursing. It’s a growing number of small gene panels, and when I say small, some are twenty genes, some are five hundred genes, but all of those are relatively small, relative to the exome. In an ideal world, you wouldn’t do that at all. In an ideal world you would do genome sequencing, because you remove the cost of selecting for those things that add a significant cost, and you find a lot more information. How do you know what you want to look for is confined to the exome, which is less than two percent of the genome? We’re starting to discover how much of the genome is not coding, and how important that is. You also can’t see structural changes in the same way and you can’t see copy number variations. With genome sequencing, if it was cheap enough, you would have the ability to make sense out of all that data. It would be the ideal way to look at these types of things.
AllSeq: An objection that we’ve heard from a number of clinicians is that they don’t know how to handle or interpret the data they’re getting back. They’re looking for the simplest report that they can wrap their brains around.
Dr. Smith: Right. And even when you send them results from small gene panels, there’s way too much information for them, information that they don’t even want. They want actionable information from the sequencing that they’ve seen. You have something like Foundation Medicine that looks at two to three hundred different genes but the reality is, there’s only a small handful of those that are actually targetable. So that actually provides way more information than the clinician needs. The ideal would be when you sequence a small gene panel, an exome or a genome, you turn around and report back not all that information but just what they need to know (to answer the question) “What can you do for this individual?” And I think it doesn’t matter how much information you generated. It’s the translation of that information and the visualization of information in a sensible manner, so the clinician can make an important clinical decision. They don’t need an encyclopedia of variants. They just want to know the three or four important things that have drugs that they could use and target it.
AllSeq: At the recent Festival of Genomics in Boston you were part of a panel on the role of non-coding RNA in cancer. Mostly we hear people talking about the genome rather than the transcriptome. How would you rate the importance of the two for clinical applications?
Dr. Smith: Well the transcriptome is extremely important. Back to what I was saying previously, in an ideal world with greater sequence output everything would be whole genome sequencing, but whole genome sequencing still doesn’t tell you what’s being actively transcribed. So there’s definitely the thought that in the future, maybe it’ll be whole genome sequencing combined with some level of transcriptome sequencing to get a more comprehensive picture. If we could make sense out of the methylome as well then I would say probably all three technologies could potentially be used. But there is no question that transcriptome sequencing in the future is going to have a place in the clinic, and there’s a lot of stuff in a transcriptome that you can look at, but there’s also a lot of stuff in a transcripome that we have no clue what it means. If you see a coding transcript, that makes more sense than if you see changes in a whole series of long, non-coding transcripts of unknown function. Ultimately with transcriptome sequencing you’re not only doing the amount of transcription, you’re also looking at the sequence of what is actually being transcribed. What do mutations, what do alterations, what do changes in expression mean? Which isoforms are being produced is a huge area and I think that’s going to have a very important clinical significance, so one cannot discount the transcriptome. There’s a tremendous amount of information that one can obtain from transcriptome sequencing.
AllSeq: Have you seen transcriptomics being adopted in the clinic?
Dr. Smith: No, but there have been discussions. Genomic Health was actually thinking of transitioning into some sort of RNA sequencing instead, but I think they’re still sticking with their platform. Right now, I haven’t seen any clinical applications of transcriptome sequencing. They’re having trouble getting reimbursement for even the small gene panels, so addin this component will complicate things. But there is no question that it’s coming.
AllSeq: What are you seeing on the data analysis side?
Dr. Smith: It’s the Wild West. Everybody is coming up with different solutions. If you’re just starting to get into this, there are so many different providers that give you different ways to analyze the data. We played with that a little and then started building our own infrastructure. Everybody is sort of on their own, and there’s nobody that’s really risen to the top. Unlike Illumina, which is the premier platform for sequencing, there’s no premier analytical platform yet.