At A Glance
Laura Reid, director of research and development, Expression Analysis
Education: PhD — Genetics, University of Wisconsin, Madison. BS — Biology, Albion College, Michigan
Employment: 2001-Present -- Expression Analysis, Durham, NC
Previous: Postdoctoral fellow and research assistant professor; director of genomics core and microarray facility, University of North Carolina,Chapel Hill.
Laura Reid is a pioneer in microarrays, not only as the founder of the genomics core facility at the University of North Carolina, but since 2001, as first the lab manager, and then, director of research and development for Expression Analysis, a 15-employee Durham, NC-based microarray services provider that uses the Affymetrix system. In the dynamic atmosphere of the maturing microarray industry, Reid’s job is to position the growing startup company for the future. She is involved in committees investigating microarray standards, helping the FDA evaluate microarray data; and, at the same time, helping EA’s customers solve research problems ranging from cancer investigation to gene-expression patterns in arabidopsis. She has processed literally thousands of microarrays in the five years she has been involved in this industry, and called on that experience in speaking with BioArray News this week about the future of microarrays and services:
How did you first get involved in microarrays?
In 1999, I became director of the genomics core and microarray facility at the University of North Carolina at Chapel Hill. I was doing a positional cloning project at UNC and it was clear that this type of work needed support from the university. One of the tools everybody wanted was microarrays. I was very lucky. I was well trained by some people from Stanford and Lance Miller, who was at NCI at the time. UNC supported me and we were awarded several grants. We bought robotics [a Gene Machines Omnigrid], and clone libraries and started making arrays for several different species using cDNA fragments and long oligonucleotides.
In the beginning, I spent a lot of nights alone with the robots. I spent a lot of time worrying about humidity, and trying to unclog the pins. There was a lot of variability going on in our home-spotting system and that was the source of many headaches. Now there are many quality microarray reagents and quality platform so that you can get reproducible and reliable results for thousands of well-characterized genes.
What made you decide to enter the commercial sector?
In 2001, I was at a point in my career that I was ready to move from an academic setting and try something commercial. Expression Analysis was a wonderful opportunity. I liked the people who were getting it started and I was ready for a challenge.
What was it like when you started and how has the lab changed?
We started with a kitchen, quite literally, a kitchen. We turned that into a laboratory with one Affymetrix system (a scanner, fluidics station, and a hybridization oven). I established the protocols, and the quality control practices, and I consulted with clients. Since then, we have rapidly grown. We are constantly renovating the lab and adding new instruments. Now we have multiple Affymetrix systems, including the old and new [Affymetrix] scanners. We use the Agilent Bioanalyzer, and have equipment for doing genotyping and RNA isolation as well. We have an elaborate LIMS system that we use to track quality control parameters generated during target preparation to identify changes in the laboratory. That really helps with the consistency of our results.
Last year, we hired Dr. Tom Goralski as our laboratory director, and we promoted Donald Cox to lab manager. We needed the lab to grow because there were more samples coming in, and we also needed our R&D efforts to have sufficient time and resources.
We started off as a commercialization of Duke University’s microarray core. Initially, we only offered standard target preparation and hybridization to Affymetrix arrays. We quickly recognized that our clients were looking to us to provide more leadership and expertise. We initiated these research and development efforts to optimize some of the protocols and develop some of our own solutions. We also became active in some standardization efforts and guidance with regulatory agencies. And, we started supporting statisticians to help with our data analysis efforts. We have several wonderful statisticians on staff. They immerse themselves in the data so now they can recognize some of the nuances in microarray data and short oligonucleotides.
What protocols are you working on now?
Some of our R&D efforts are looking for protocols in preparing targets from limited or difficult RNA sources such as blood, or formalin-fixed paraffin-embedded tissues. We have had good luck there and are developing some proprietary protocols.
Let’s talk about your work in R&D.
R&D in a small company may be different than in a big company. I’m involved in very immediate needs. For instance, we recently had a client with 100 samples that required RNA isolation from stomach tissue. Since we didn’t have an established protocol in place, R&D was responsible for developing an extraction procedure for stomach tissue. I’m involved in the External RNA Controls Consortium, which is a group of about 30 representatives from private and public and academic sites who are working together to produce external RNA spike-in controls that can be used to monitor gene-expression performance on a variety of microarray platforms and by RT PCR.
Also, what we have noticed is that, although everybody uses very similar protocols, minor differences in your procedures, instrumentation, and analysis methods can affect your microarray results and that can affect your data comparability. So, Expression Analysis is starting a proficiency-testing program, which involves collecting data from multiple laboratories, using the same RNA sources. We plan to collect it over a year, with multiple rounds of testing. We are looking for the reproducibility within labs and between labs, and also what we call data comparability — that is, did all labs identify the same differentially expressed genes?
At this point, there is no one gold standard, and our survey should provide a broad distribution of microarray results. We envision a proficiency testing requirement for labs in the future, particulary those submitting data for regulatory review. We hope that some of the data we are generating now can be used to help estalbish those guidelines.
How is the RNA consortium working? It seems like something like a goal of recommending an RNA standard could be accomplished in a year.
You would think so. But what turns out is that Affymetrix users are using one set of controls — that’s one set of spikes. People making their own spotted arrays are using a different set of spikes. Stratagene is selling a third set of spikes. There is no consistency across the platforms. There is also no consistency in how you analyze this data. So, part of the initiative of the ERCC is not only to say: ‘Here are common transcripts that will be used on all platforms and in RT-PCR,’ but also to set up software guidelines to say: ‘Here is how we are comparing sensitivity and specificity.’
This has now morphed into an NCCLS initiative [a standards advocacy body within the clinical laboratory community]. Janet Warrington at Affymetrix along with Marc Salit at NIST sent in a proposal to NCCLS, which was accepted. And, now a committee is reviewing these external RNA controls.
When do you think things will come about that will move the field forward?
The ERCC is working to have a product out by the end of the year. That’s still kind of tenuous.
The microarray market has moved very quickly toward maturation as an industry. What are potential customers saying now?
In the beginning, you would run into people who didn’t believe microarray data, or felt you should only be looking at protein expression. I don’t run into them as often. I think there has been a lot of good work showing how reproducible and reliable the results are, and how incredibly useful they are. Now, microarrays are being used to identify pathways in a discovery setting; they also are being used to predict toxic affects or assist in disease diagnosis, and someday to distinguish patient responses. So I think it is becoming clear that RNA expression profiles are a part of drug development efforts and they eventually may be a standard tool in clinical settings.
What is the future for academic core labs?
We feel that several academic and government research centers will elect to outsource their microarray services. We are Duke’s service provider, and we have major contracts with other academic and government facilities, including the EPA. I think outsourcing microarrays will become more common. Some people are outsourcing because their internal facilities are overloaded, so we get that runoff as well. Beyond that, what are we looking for in the future? There is this transition to the clinical setting. We have made all of our laboratory practices and all of documentation GLP compliant so that the data generated here can be submitted to regulatory agencies. We also collaborated with members of Schering-Plough on a mock submission to the FDA. That started last May, and there were several drafts over the summer, but the final submission was in late October.
What has been the feedback you have received from that process?
We have gotten lots of very good feedback. As I understand it, the FDA received six submissions of microarray data, some of mock [data] and some real, and in many different formats. We got lots of feedback on what quality controls metrics they would like to see, how data should be submitted in the future. It was an excellent experience and we continue to talk to the FDA and work together.
Later this month, the FDA will be presenting a symposium, it’s called the Science Forum [www.dcscienceforum.org], on May 18 and 19. And we will be presenting a poster there. The members of the pharmacogenomics subcommittee are presenting a poster there talking about all these microarray submission experiences. It is my understanding that they are heading towards another guidance document. What we wanted to know from them is, which quality controls are they going to be looking at? And what they wanted to know from us is what is a good range of data that they can establish for quality control parameters or ranges. I think they are still working on that.
You are specialists in the Affymetrix platform. Are there thoughts about the need to add deep expertise in a second platform and what are the advantages and disadvantages to doing that?
That is a complicated question. There are obviously scientific issues and business issues involved. We have been very happy with the Affymetrix platform, but we recognize that some people prefer other methods. Clearly Affy is market leader — I think something like 80 percent of the pharmaceutical market uses the Affymetrix platform. And, it certainly is an established leader. In the long term, it is not clear whether people will be using smaller focused arrays, or alternative technologies to look at multiple analytes at the same time. We are evaluating all these platforms, but at this time, we are sticking with Affymetrix.
Santa Clara will appreciate that. What are your thoughts on the emerging next layer of technology: You do your gene-expression profiling, and then you go to a next set of instruments?
What would be lovely would be if you could complement the RNA services with protein services. I’m sure in the future, people will find a protein biomarker that works best on one specific disease, or other people will find that RNA expression profiles are better at predicting a certain drug’s responses. In the future, we are probably going to need a variety of tools.
How far away do you see that?
The protein field is still in development. So, I doubt that Expression Analysis would transition for, say another three years?
What is the biggest shortcoming of the technology as it is now, and what are the easy fixes?
We need more standardization. We need to know that the results you are getting on this platform are comparable to the results on that platform, whether the results you get from that lab are directly comparable to the results you get from that lab. At Expression Analysis, the way we think the [of] the road to standardization is through some sort of data comparability and proficiency testing program.