Outsourcing your RNA

The Scientist spoke to researchers about what motivated their outsourcing decision, and what others can learn from their experience. Here’s what they said.

Researcher: Carl Bruder, advanced research scientist, Southern Research Institute, Birmingham, Ala.

Project: De novo sequencing of the ferret transcriptome²

Problem: For de novo transcriptome sequencing—that is, without a genome scaffold on which to align reads—cDNA normalization is “of uttermost importance,” says Bruder; the step ensures that all transcripts are picked up, and not just the most highly abundant ones. But, it is tricky to do properly, Bruder says, and not something he felt comfortable taking on.

Solution: Bruder knew what steps he needed to take, but not how to go about performing them. So, he made sure to pick a service provider that did.

“I didn’t feel really happy about sending a sequencing job worth $20,000 to a lab that did not know how to normalize a cDNA library,” he says, “so that was one of the criteria when we selected an outsourcing company.”

Bruder kept that criterion to himself, however, and waited for prospective companies to bring it up. “I wanted the service provider to ask me whether I wanted [the library] normalized or not,” he says. “Those that never thought about that were never on the list of companies to choose from.”

In all, Bruder interviewed some 10 companies and core facilities before finally settling on MWG Eurofins, based in Munich, Germany, and Huntsville, Ala., which sequenced his samples using Roche’s 454 chemistry.

The local connection was important, Bruder says, but not a deal breaker. He got a bit of an “Alabama discount,” and the proximity meant he could meet company representatives face to face. But as the company’s sequencing facility is in Munich, most of their scientific interactions actually were by teleconference.

“I am quite happy with the service I bought,” Bruder concludes. “I got a lot of value, and will use them again.”

Considerations: Bruder advises would-be outsourcers to do their homework. Rather than simply farming out a job and trusting the technicians to know what to do, read up on the technique, and consider sample preparation and experimental design accordingly. “Had I just gone with the first or cheapest sequencing core facility, I may have been dissatisfied with the results,” he says.

Researcher: Belete Teferedegne, Oak Ridge Institute for Science and Education fellow, US Food and Drug Administration

Project: Identification of miRNA biomarkers of tumorigenesis

Problem: Biomarker discovery means microarrays and qRT-PCR, and that requires both equipment and money. Teferedegne’s lab lacked the former, but had the latter. Yet with PCR assays running into the hundreds of dollars apiece in-house, he says, it simply made financial sense to farm that work out.

Solution: The miRNA array work, Teferedegne says, had to be outsourced; the lab he works in lacks the hardware to perform the necessary hybridization and imaging. He could have handled the subsequent validation—confirming individual transcripts expression levels via qRT-PCR—“but if you want to run 10 or 20 miRNA assays on 20 or so samples, it is very expensive.” The reagents for each assay cost $250 to $300, an investment that’s negligible when processing hundreds of samples, but makes little sense if any given assay will only be run once or twice. Plus, there’s the inevitable optimization and troubleshooting to consider.

So, Teferedegne outsourced his work to LC Sciences, a Houston-based service provider, which builds and uses custom miRNA microarrays. The decision, he says, came down to communication. “They were very good,” he says. “I didn’t know much about microarray work then, and [the company] helped me to make a decision and determine which platform was good for me.”

Using total RNA samples supplied by Teferedegne and an array of some 750 primate microRNAs, the company helped him identify 90 or so transcripts that appeared promising. Of those, he chose 10 for validation via qRT-PCR, to ensure that observed expression changes could be independently observed. He then selected three to five miRNAs for in-depth analysis using qRT-PCR in his own lab.

Teferedegne says LC Sciences charged $795 per chip (including preliminary data analysis) for the array work, but just $30 per qRT-PCR reaction, turning samples around in 4 to 6 weeks. The overall experience, he says, was “excellent.”

“They are a little more expensive than others, but it’s worth it,” he says.

Considerations: If you’re going to outsource any sort of big data research, especially microarrays or next-gen sequencing, don’t forget the downstream bioinformatics. Firms like LC Sciences generally provide some basic data crunching as part of the service package, and may offer more comprehensive analyses for an extra fee. In Teferedegne’s case, that assistance, in the form of discussions with his company representative, gave him the confidence to handle subsequent bioinformatics work on his own. “I’m not trained as a bioinformaticist. I needed someone to help me the first time, at least,” he says. “Now I can do it myself.”

Researcher: Wolfgang M. Schmidt, university assistant, Medical University of Vienna

Project: miRNA profiling to identify transcripts involved in the innate immune response³

Problem: Schmidt is an expert in using Affymetrix GeneChip microarrays, but at the time of this work, the company wasn’t marketing a microRNA array. Another company, Exiqon, was, but “we didn’t have the instruments needed to conduct the experiments,” says Schmidt.

Solution: He knew where he could find them, however: A friend who recently had become director of the Centre for Proteomic and Genomic Research (CPGR) in South Africa told Schmidt he had the Exiqon platform up and running in his Cape Town facility.

“I knew the CPGR had set up state-of-the-art hybridization equipment and the data would be of top quality,” Schmidt says. And that was important: Having managed his own GeneChip facility in Vienna, and watching colleagues struggle with homemade or “inferior” equipment, Schmidt knew first-hand how critical infrastructure can be.

For about $650 per sample, Schmidt submitted total RNA, and “exactly 3 weeks later” was processing the raw array data himself.

Bioinformatics, he insists, should be handled by the investigator; only they will understand the nuances of the biological questions being asked and thus how best to extract meaning from the data.

“You need a lot of face-to-face interaction with a person who can do this,” he says. “I am not convinced a service provider in Cape Town or in Germany or in the US can do this, if you don’t have the possibility for a direct personal interaction.”

Considerations: If you don’t have a personal connection, says Schmidt, just ask the vendor of the hardware platform you will be using (in his case, Exiqon). Many companies maintain lists of reputable service providers and will provide them if asked; some even publish them on their Web sites. When Schmidt ran his GeneChip facility, he says, Affymetrix routinely contacted him to evaluate how well his team was doing. “So I would advise you to ask the people who are selling the chips, because they are interested in sending you to a service provider who is performing really well,” he says.

Researcher: Zhi-Ming Zheng, senior investigator, National Cancer Institute, NIH

Project: MicroRNA screening in normal and cancerous cervical tissue⁴

Problem: At the time of this study (2006), the NCI did not have a microRNA array service. Plus, space is at a premium at the NIH—“We use a ruler to measure which inch is yours and which is mine,” Zheng quips—so setting up his own array facility was out of the question.

Solution: Even if space were no issue, says Zheng, he still wouldn’t want to go to the trouble of establishing a microRNA profiling system in his lab. Getting a microarray workflow up and running requires a serious investment of time and resources in terms of standardization, quality control, and troubleshooting.

“I think I would rather have [a service provider] do it,” he says, “They have more experience with troubleshooting issues, and more skill with the bioinformatics.”

Fortunately, companies were by then beginning to offer miRNA array services, and Zheng “took a gamble” on one of them, LC Sciences. For “maybe a few hundred dollars” per sample, he submits total RNAs and two weeks later has the data, processed and ready to use. “That’s very unusual,” he says; core facilities at NIH are so overworked that turnaround times can be measured in months, he says. “By the time you get the data, you’re already doing something else.”

In this case, says Zheng, the resulting data were of very high quality; hit testing by Northern blotting yielded almost 100% validation.

Considerations: Since this study was completed, NIH has established its own miRNA array facility. Yet because of Zheng’s positive experience with LC Sciences, he continues to outsource his array work, and even recommends the company to his colleagues—an important source of information when selecting a service provider. Just as with any major purchase, word of mouth is key. “I’m sure LC Sciences got a lot of business from the NIH,” he says.