While genome-wide RNAi-based screens remain a popular gene-discovery tool, an increasing number of researchers are using the technology in a more focused manner, often in conjunction other genomic approaches.
Though the data that result from these kinds of narrower screens tend to be more manageable than those that are generated from genome-scale ones, both kinds of studies present similar pitfalls that can stymie investigators' efforts to make sense of their findings.
According to Stephanie Mohr, director of the Drosophila RNAi Screening Center at Harvard Medical School, as the research community has grown more familiar with RNAi, scientists have come to understand that genome-wide screens are going to produce a massive amount of data.
"That becomes quite a big project to follow up on," she said. Increasingly, "we're seeing … people who have done some kind of other large-scale project — a large-scale proteomics study, say — come to us to follow up on [a subset of hits] to get some type of functional data from RNAi screening."
So Young Kim, director of the RNAi Screening Facility at Duke University Medical Center, has also had customers using RNAi to follow up on results produced by other technologies. "Using lots of other genomic technologies, people have generated interesting lists of candidates, either through copy-number alterations or expression analysis or deep sequencing … that they want to functionally interrogate," she said. "The next logical step is to translate that into a custom or small-scale RNAi screen."
Michael Green, director of the University of Massachusetts Medical School's RNAi Reagent Core Facility, has also noticed an uptick in the number of people interested in focused library screens rather than genome-wide ones, in part because they tend to be concentrating on a particular biological problem and therefore already "have a sense of what kinds of proteins and genes they're interested in studying."
Narrower screens are also technically easier and cheaper, he noted, since they require a smaller investment, both in time and reagents.
Indeed, the expense of a genome-wide RNAi screen can often prove too much for investigators, especially when they rely on funding from government agencies like the US National Institutes of Health.
Grant proposals for genome-wide screens have traditionally faced an uphill climb, Green said, because NIH prefers to support "hypothesis-driven projects" and views such screens as "fishing expeditions."
Kim agreed that funding such projects can be a challenge. "It's difficult to get money from the NIH to do large-scale screens," she said, "because there is no guarantee that you're going to get some sort of conclusive result."
But even with a small, focused RNAi screen, obtaining useful results can be a challenge.
"You're going to get a lot of hits, and you can't follow them all," Green said. "So you have to pick the ones that you think are going to be the most interesting," by focusing on ones that appear to have the most clinical relevance, for instance.
"Most [screeners] don't know what they're getting into," he added. "I think it's important to have a strategy of how you're going to follow up hits, because you're going to get hits from the screen — they're all designed for that."
Harvard's Mohr said that she always encourages investigators to prepare a data-analysis plan ahead of time.
For example, image-based assays can yield a "tremendous, terabyte-sized dataset," she said, "and one is not going to look by eye at a genome-scale collection of images. You need to collaborate with someone who is an expert or, at least, use some sort of automated software tool to approach that problem."
She also said that it is important for researchers to understand and be ready to address the limitations of an RNAi screen.
"People need to be aware that some RNAi reagents … are going to compromise cell viability," she said. "So cells might be a little unhappy, and you have to make sure you're accounting for that in your assay readout.
"You have to make sure, in the same way one did with classical genetic screens, that you'll be able to see what you're interested in, separate that from the wild-type phenotype, and make sure that it's really giving you the set of genes you're interested in," she added.
A potential complication that threatens every RNAi screen is the less-than-perfect reliability of gene-silencing reagents, Kim noted.
"Some people express surprise that the library they are working with isn't 100 percent validated," she said. "You have to introduce them to the rough reality."
An important consideration for screeners is to incorporate this uncertainty into their data analysis so that they can eventually end up with a final list of high-confidence hits.
"Everyone approaches this question a little bit differently, but it is something that every screener has to think about … [to] come up with a good solution that works for them," she said. "These are details that people don't like to mention, but a big part of doing an RNAi screen is working your way around them."