Setting up a new gene expression lab is a daunting task. More and more, though, it is falling into the domain of individual researchers, not just large groups or core facilities.
"With genomics or functional genomics technology reaching a very mature point, I think it's something that individual labs can tackle now," says Penn State's Willard Freeman, a neuroscientist and director of the functional genomics core facility there. "It isn't a specialty that's off on its own. If you're a cancer lab and you do cancer research, a lot of tools can be brought right into your own lab and you don't have to spend years getting stuff up and running. It's almost to the point of plug and play."
Choosing those tools, though, takes a bit of thought. For gene expression analysis, the two most common methods to determine changes in expression are microarrays or quantitative PCR. While those methods remain popular and useful, other technologies, such as next-generation sequencing, are horning in on their territory. However, those big-ticket items aren't the only things needed to keep a gene expression lab running smoothly. Just as vital are tools for quality control and data analysis, as well as knowing what to avoid.
The most obvious item a gene expression lab needs is a way of measuring said expression. While new tools and methods are becoming more common — we're talking about you, sequencing — microarrays and real-time PCR are still in full-swing use.
Generally, microarrays (and increasingly, sequencers) are used for discovery stages. "Normally, what people do is if they are interested in a bunch of genes, they will do a microarray study and then what they will usually do is validate results with real-time PCR," says Steffney Rought, manager of the genomics core at the University of California, San Diego's Center for AIDS Research. "Unless they know exactly what they are looking for, they are going to do a microarray."
Wei Wang's core at Cornell University has just about every microarray platform and every level, including Affymetrix, Illumina, and NimbleGen. He adds that if he had to choose just one array platform, he'd go with Agilent's. "I think the instrument itself is not that expensive," Wang says. "It's basically a big microarray scanner. All the accessories are minimal cost."
At the Jackson Laboratory in Maine, Doug Hinerfeld's expression core handles both arrays and qPCR, though he says the lab is microarray-centric. There, they study mainly mice and use Affymetrix Mouse Gene 1.0 ST arrays, but he adds that the choice "depends very much on what organisms you are interested in."
Affymetrix's Rebecca Brandes, an associate director of product marketing, says that use of the company's GeneChip Gene 1.0 ST and Exon 1.0 ST Arrays is growing. These arrays, she adds, use probes found along the length of the gene, and the exon array has 40 probes per gene and four probes per exon. Affy also expects to launch its GeneAtlas System soon, targeting lower-throughput labs.
At NimbleGen, the most popular arrays used by academics are its 385K and 4x72K format, according to Rohaizah James, the product manager for expression. She adds that prokaryotic-focused researchers prefer the 4x72K array and eukaryotic-focused people tend to go for the 385K array. "What we have been working for in the last couple of years is increasing our multiplex format to allow for more convenient high-volume analysis," James says.
Smaller labs generally don't need to have the most cutting-edge equipment or even established tools that would only be used infrequently. "For the individual investigator's lab, I generally don't think that the very large equipment — Affymetrix readers, Illumina BeadStations, or the next-generation sequencers — [is useful], unless it's a very, very large group," Freeman says, adding that few labs can do a microarray study enough times a week for it to be worth the cost.
Choosing which tool to bring into your lab is a matter of your research focus as well as your lab's size. "I would argue for individual investigators to probably focus on an instrument, whether it be a real-time PCR instrument or some of the new technologies — Nanostring sort of replaces qPCR — because you are going to do a lot of these studies," Freeman says.
Rought's genomics core at the Center for AIDS Research, which she calls "higher throughput [but] not super-high throughput," has the ABI 7900 platform that can handle both 96- and 384-well plates as well as low-density TaqMan arrays and Roche's LightCycler, which she says are "pretty common real-time machines."
Freeman's functional genomics core offers both qPCR and microarray services and is testing out new technologies for the middle-throughput area. "That middle ground is the area of the greatest growth at the moment," he says. "I think that's where most of the new technology is coming out: qPCR fluidic chips from ABI or Fluidigm or BioTrove or things like Nanostring, which is a completely different technology that allows you to multiplex your gene expression work."
BioTrove's OpenArray combines aspects of arrays and aspects of microplate technology from PCR, says company vice president Kevin Munnelly. "We use solution-phase technologies for real-time PCR, like TaqMan and SYBR, but we're in a parallel format, which is very similar to the arrays," he says. The OpenArray plate has 3,072 holes in which any configuration of primers and probes can be placed, giving the user 3,000 to 9,000 datapoints. Munnelly adds that this platform is for use after you've gone through a discovery step.
Nanostring's nCounter System also comes in downstream of either microarrays or next-generation sequencing. This tool uses two probes, a capture and a report probe, to determine what genes are expressed and by how much. "Customers can multiplex up to 550 gene targets in a single assay," says Lianne McLean, vice president of marketing. "One of the real performance benefits of our platform is they can multiplex and they can get very highly accurate and highly sensitive results."
Freeman's lab is currently comparing Nanostring's nCounter to Illumina's BeadExpress and to traditional qPCR. "All of them work quite well," Freeman says. He and his lab are still working on writing up their results.
Peter 't Hoen and his colleagues at Leiden University Medical Center have begun to offer next-gen sequencing in addition to their microarray and other services at the Leiden Genomic Technology Center. Last year, 't Hoen and his colleagues published a study in Nucleic Acids Research showing that the digital gene expression assay from Illumina was more robust than either microarrays or qPCR.
With DGE, the researchers were able to see more differences in gene expression between hippocampal regions of wildtype and transgenic mice than they could with microarrays or even qPCR. "This platform has more functionalities and more power and lot of other advantages over microarrays," says 't Hoen, adding, "To make a bold statement, this next-generation sequencing platform is actually more accurate than quantitative PCR." If he were to start a gene expression lab from scratch today, 't Hoen says, he'd skip right to next-gen sequencing.
Hitachi Central Research Laboratory's Hideki Kambara agrees. "For the moment, I would like to use a massive parallel DNA sequencer for gene expression analysis because it is very accurate, although it is very expensive," he says.
Hinerfeld's core, which he classifies as medium-throughput, is also investing in a next-gen sequencer from Illumina. "High-throughput sequencing is where all the energy is focused. It seems that pretty much all of our customers now when they talk about gene expression analysis whether it's microRNA or mRNA, they are also saying, 'Well, should we be doing this with high-throughput sequencing?' And it just comes down to price," Hinerfeld says
Not just the big tickets
A good, well-run gene expression lab has other considerations beyond the old microarray versus qPCR, and now versus sequencing, debate. A smoothly running lab also has to have quality control checks, well-established standard operating procedures, and a way to handle the data coming off the runs.
Nearly everybody agrees on quality control. When dealing with nucleic acids, you want to be sure you've isolated enough to be used in your study. For isolating DNA or RNA, there are tissue homogenizers and tools for extraction available, but depending on the lab's needs, manual work could also be the way to go. UCSD's Rought has a QiaCube that she uses for nucleic acid extractions, but Jackson's Hinerfeld prefers to go it manually, especially as his lab handles a variety of tissue types.
To make sure the nucleic acid yield is of high enough quantity and quality, researchers often turn to the NanoDrop, Agilent's Bioanalyzer, or to a spectrophotomer for the OD 260/280 — experts say this is not something to skimp on. "If you are going to do a lot of this, I would go ahead and recreate that in your own lab. You'll use it enough and it's not too expensive," Freeman says.
Even with the best tools and quality nucleic acids, the instruments have to be used properly. Having standard operating procedures for the process is crucial, Hinerfeld says. His lab has developed protocols over time and to the point where, for an experiment with 800 microarrays, they cannot see any differences between the multiple hands processing the samples.
Rought concurs, adding that it is crucial to have good people. "There are so many ways that you can mess up the gene expression experiment," she says. "Having competent people is just as important as the equipment you're purchasing."
There's also the matter of keeping track of the samples, particularly for larger or higher-throughput labs. Hinerfeld says that a LIMS systems has been useful for his lab and has allowed them to go back and query past sample runs. Their LIMS also sends the raw data on to the bioinformatics group at the Jackson Lab for analysis.
Not every institution has such a bioinformatics group, so a gene expression lab may have to handle the data itself. Before even buying that microarray scanner or real-time PCR machine, Freeman advises people to take a look around their lab and make up a plan for their computer infrastructure. "This doesn't have to be supercomputing or anything, but [have] a plan for dealing with the number of computers that you end up having in a lab collecting data," he says.
There are some items, though, that most gene expression labs can avoid without any ill effects on their work. Much of it is not being tempted by the lure of a shiny new toy that you'll barely use or tools that aren't appropriate to your level of throughput. "We try to avoid buying those things that are super-gimmicky and all the rage and you end up not really using it," Rought says.
At her lab, the one item they bought but wound up never using was a robotic system that prepared and even loaded samples. "[Researchers] get a grant and they think, 'Oh, yeah, I'm going to do all this qPCR' and it turns out that something will go wrong and their experiment won't go as planned and that's one thing that can definitely not be used," she says, adding that such systems are more likely to be necessary in industry and biotech than in academia.
Freeman, however, would caution against purchasing a 96-well qPCR machine and instead says that 384-well machines give much more data and use fewer amounts of reagents. "That savings in reagents really adds up. It adds up for us to tens of thousands of dollars a year. We have a 96-well PCR machine that we are going to retire and everything that we have will be 384-well," he says.