Name: Tina Hambuch
Title: Senior Scientist, Illumina
Experience: Scientist, Ambry Genetics; Assistant Professor at Ludwig-Maximillians University of Munich; Post Doctoral fellow, Centers for Disease Control and Prevention
Education: PhD in genetics, University of California, Berkeley; BS in biology, University of California, Irvine
Illumina began offering whole-genome sequencing within a CLIA-certified and CAP-accredited laboratory in 2009.
Senior scientist Tina Hambuch serves as the company's scientific liaison to the clinical laboratory where she works to develop clinical tests based on next-gen sequencing, design validations for those clinical tests, and develop analytical tools for whole-genome sequencing.
Last month, Clinical Sequencing News visited Illumina's headquarters in San Diego, took a tour of its CLIA laboratory, and spoke with Hambuch about clinical sequencing's future as well as the process for sequencing under CLIA guidelines and how it differs from sequencing for research purposes.
An accompanying video of the facility tour is available on GenomeWeb's video channel here.
Can you describe the process of sequencing whole genomes within your CLIA lab, and how that differs from sequencing for research purposes?
Because it's a clinical lab, it has to start and end with the physician. One big difference is that the physician typically is ordering this for a specific reason, so we at the clinical lab have to be ready to help support the physician in understanding exactly what he or she is going to get from the product.
For example, if this were a simple, straightforward cystic fibrosis test, you want to be able to have support in place to tell the physician, this is what we cover, this is what we don't cover, this is how it relates to other tests that might be out there that might help you resolve the questions you're asking about your patient.
So one of the first things is just having the right support and educational material in place to be able to help physicians to decide when this is appropriate. We don't want anyone to order a test for something that isn't going to help them resolve their question, or at least have that potential.
That's the first step. And in order to help with that, we have a variety of materials that we've developed that include things like an informed consent process. A physician and the patient have to get together and discuss what they can learn, what they want to learn, what maybe they don't want to learn, what they're willing to look at, what they aren't willing or able to look at.
Before going through the actual ordering of the test, there's a lot of up-front preparation and counseling that happens both between Illumina and the physician, but also between the physician and/or genetic counselor and the patient to try to understand how the information can and should be used, and to make sure that everybody's in agreement about how it should be used going forward.
So, in some cases you're not returning all the results of the sequencing?
Exactly. A lot of times physicians are trying to ask a very specific question and certainly if it's a child there are guidelines out there that say that certain genes shouldn't necessarily be looked at in children because it's an adult-onset condition and it raises a lot of ethical dilemmas. It's up to the doctor and the patient to make those types of decisions together, and they're going to be different for different people and different situations.
Throughout the course of offering clinical whole-genome sequencing, how has your informed consent procedure evolved? Is there anything you've had to add because of unanticipated consequences?
It's actually gotten simpler. When we first put this out, we tried to cover every possible contingency and we had this nine-page consent form that doctors were finding a bit too onerous. We've moved toward a simpler consent form that looks more like consent forms that you would sign for any other genetic test.
What's the process of bringing technology into the CLIA lab? How is it validated?
The main focus in a CLIA lab, aside from being able to provide the physician support, is making sure that the test meets certain minimum standards and has a certain minimum level of reliability for any sample.
There are a number of things that we try to evaluate. We run a lot of samples and we measure all of the quality metrics we can think of around those samples and try to establish cut-offs to say, when the quality metrics are performing at this level or above we always have a good outcome, and below that, we don't necessarily trust it.
The main thing is going through a validation process. That breaks down into two kinds of questions. One is: if you get an answer, is your answer correct? And if your answer is negative, how confident are you that you didn't miss something?
Obviously, it's not necessarily feasible to run thousands of genomes, or even hundreds of genomes, for validation, [so] we took this through validation using a multiple tiered approach where we asked very specific questions of known clinical samples.
Then we also took a few genomes through whole-genome [sequencing] and did comparisons to other orthologous technologies that are measuring genetic diversity at a genomic level. [For example], we took our microarray assay that measures over two million variants distributed across the whole genome and asked, 'How often were the calls made on that technology concordant with the calls made on our sequencing technology?'
Is that the same process for validating reagents?
Yes. Every time you get a new reagent, even if it's the same kind of reagent, you have to do sort of a mini-validation to make sure that that Taq polymerase, or whatever it is, is performing at the same level as the previous one. There are mini-validations that happen on a daily basis in the lab.
In addition to that, if you have a major change like a chemical change or a bioinformatics change, you have to go through a big validation again. One thing with the next-gen sequencing technology is that it does evolve quickly, so we've done many of those bigger validations already, which is also great because it's helping us feel more comfortable about knowing what we're measuring.
Is there a difference in turnaround time when doing sequencing in a CLIA lab, versus sequencing for research?
It's probably in practice about the same. I think in terms of what we promise, it's a little bit longer. On the research side, you order a test or you order service, and in our lab things are run at high quality and there are quality metrics and measures, but there aren't necessarily the same thresholds applied. If you apply those thresholds [for clinical sequencing] sometimes you have to run additional samples or additional lanes in order to meet the same quality standards that you're promising. So it can take a little bit longer sometimes. Then the back-end [quality control] process can be a lot more laborious.
How does the back-end process differ?
After sequencing … it has to be analyzed bioinformatically. A whole new set of filters and QC steps apply where data are trimmed out based on their performance. The performance and the trimming qualities that we establish are based on a lot of the metrics that were established through the validations. So, in our validations we became 99 percent confident of a call when it met these set of six criteria and, based on that, all genomes going forward have to meet these criteria for the calls that we can make. So all of that trimming and processing has to happen.
Once it goes through that, it has to be manually evaluated. Somebody actually has to go through and look at the overall statistics and compare those to normal genomes, genomes we know have been good in the past, and make sure that we're not seeing any kind of unusual trends.
We sometimes do see something that's really unusual if somebody is of a different ethnic background, for example. One of our first samples was from a person who had a very mixed background, and their diversity was completely off the chart from all the other genomes we'd ever done. We actually spent quite a lot of time making sure [the sequencing results were] in fact real and not some sort of strange, inaccurate outcome.
We do a lot of things like that and we double check that … making sure certain things about the patient, [whether] they're male or female, [for example,] makes sense and matches what we're seeing. We go back and try to make sure as much as possible that the data we're looking at makes sense relative to the sample that we know we got.
At the end of that process, our laboratory director has to go over everything that has been generated and then he has to sign off on the report.
What have been some of the major challenges in bringing sequencing technology into the CLIA lab?
I worked at a CLIA lab that did genetic testing using Sanger technology before I came here, so from a sequencing standpoint I felt pretty comfortable. But [transitioning from] one kind of sequencing information to a new technology, … it's just so much bigger. There's just so much more to think about.
In particular, I think the fact that when you're looking at only one little piece of the genome, the characteristics within that one little piece tend to be pretty consistent, even if it's across a big gene. Whereas, across the whole genome, you have to be aware of how the technology performs differently in different regions. High GC regions, low GC regions, regions with homopolymers or repeats are going to behave differently.
How do you deal with that fact that sequencing technology works differently depending on where you are in the genome? Is it a bioinformatics fix, improvements to the sequencing technology itself, or by validating with another technology?
We do global evaluations using our microarrays. Then we also go in and very specifically take clinical samples that have certain characteristics. We pull up things that have tri-nucleotide repeats and other things like that and we sequence very, very deeply on those samples around those regions, and we see what we need to do in order to make the correct call in those regions and how different the quality metrics and analysis need to be in order to call quality there relative to some of the other places where we already know how to make the calls.
How do you see sequencing within CLIA moving forward and growing?
Well, you're talking to the converted, of course. I think it's going to be really great. I really do think that where we are now is so different from where we were three years ago when we first launched [clinical whole-genome sequencing]. The biggest challenge is still in the interpretation and management of all that information.
I think over time, we're going to figure out better ways to make sure we can really capitalize on all the value that's there and make sure we can ask questions over time — so as people have questions, they will be able to access this information and get better interpretations.
The other thing that's going to happen that will really enable this, is we will get more information. The more we sequence, the more we'll discover. In addition to being able to help resolve some of the questions that are big concerns right now like variants of unknown significance — we will start understanding those variants — we'll also be able to correct some things that are incorrect.
There are a lot of rare variants out there that have been reported as probably associated with conditions and people have been making medical decisions based on the limited evidence that's out there, and hopefully we'll be able to provide more evidence to help resolve those questions.