This story originally ran on March 25.
By Tony Fong
Though proteomics core facilities face a gamut of challenges, the predominant problem — one without a consensus solution — is poor sample submissions by clients.
At least that was the predominant view of a roundtable discussion at the annual conference of Association of Biomolecular Resource Facilities held in Sacramento, Calif., this week.
From questions about integrating the newest technology into their facilities to handling unrealistic expectations from clients, proteomics labs have no shortage of issues that can result in compromised data. But judging from the panel and comments from the audience, one overrides all other: sample submissions.
In some instances, a client will go to the lab with a sample that is of such poor quality that the lab doesn't know what to do with it. James Farmar, assistant director of the mass spectrometry core at University of Virginia Health System, said that his facility has seen tubes come in "with strange colors, not only of proteins. … We've seen them contain broken-off pipette tips."
While it's obvious that "something is not being done right by the supplier of the samples," he asked what the core's role should be when such samples come in, and how much effort it should devote to educate the client about what he or she should be doing.
Other audience members and panelists raised examples of clients bringing in samples that their facilities may have little or no experience handling. Brett Phinney, who runs the proteomics core facility at the University of California, Davis, said his lab received a request to look for different protein expressions in cranberries.
"I don't know how to prep for cranberries, and a lot of things are really different in plants" compared with humans or animals, which his lab more typically handles, he said.
And sometimes the samples being submitted are so unusual that it leaves core staff scratching their heads. Larry Dangott, director of the protein chemistry lab at Texas A&M, recounted that one client wanted his facility to run samples of lion urine.
"Trying to collect urine from a lion is not easy," he joked.
Turn Them Away?
There was no consensus on how to handle such requests, with some advocating turning away clients who submit questionable samples. John Asara, director of the mass-spec core at Beth Israel Deaconess Medical Center, said that he won't allow anything into his lab "unless it's from a source that I can deal with, so if it's not on a gel or if it's not in solution and has to be sort of purified by immunoprecipitation, I won't touch it.
"So I never want to see a cranberry, I never want to see a brain tumor," he said. "By the time I get it, it always has to be in a form that I can actually deal with."
Others were not so sure that such a course was not short-sighted or even feasible, with some audience members saying that running poor samples has value even if the result is poor data or no data.
Melissa Sondej, a staff scientist at the W.M. Keck Proteomics Center at the University of California, Los Angeles, said that on one project, a client presented a 1D gel sample with a "smear down the whole gel." Even though she cut out the band, "you're not going to see what's really in the band; you're going to see everything from the smear."
Nonetheless, she ran the sample, and "it was junk in, junk out." But, she added, the experience taught the client to improve his own skills and present better samples for future analyses. "So, yes, it was junk in, junk out, but it was also a good learning experience for them, so sometimes you might want to run some of that junk," she said.
Dangott at Texas A&M said that one strategy that has worked for his facility is to offer workshops where attendees may provide samples that his lab generally doesn't see. In one recent workshop, participants presented a citrus fruit and a potato, things Dangott said he had never worked with.
But because the participants had extensive experience with the organisms and could offer their expertise, Dangott was able to "come up with some nice extraction techniques," something that would not have happened if the samples were given to him outside of the workshop environment, Dangott said.
Phinney, pointed out that turning away samples may carry serious consequences: a client may complain to higher-ups.
[ pagebreak ]
"I think that's a reality for a lot of campuses," he said.
Mike Myers, a group leader in the Protein Networks group at the International Centre for Genetic Engineering and Biotechnology in Italy, said that if a lab runs a questionable sample for analysis, it's important to get copies of the results to the top of the food chain at the facility, "especially if you're writing a summary analysis about what you personally thought about the sample and the results and how it can be improved, or how it should be interpreted."
Even if the client was told from the start that the sample wouldn't work, "they may not tell their boss that it didn't work because it was something they did, so unless all of those things are going to the guy who pays the bill, you're probably not going to be able to fix the problem."
New Technology is Great, But ...
In addition to sampling, implementing new technologies in the lab was identified as a major challenge during the session. In his presentation, UC-Davis' Phinney said that incorporating the newest instruments and methods can involve numerous runs or extensive data analysis. That will cost "a lot of money," which clients may balk at, he said.
For example, targeted proteomics has become a major research strategy and various methods have been developed to do such analyses. His lab does some targeted proteomics, Phinney said, but "we found it very, very difficult to fit that into the core facility model."
One talk at the conference this week covered an iterative search strategy to optimize the transitions. While he said the method is "a wonderful idea, I'm sure it really works great, the problem I have is when I try to tell people that, 'Your sample is not going to work the first time, it may not work the second time, but we have this iterative process, [so] probably by the 10th time we run your sample we may get some data to actually go out and analyze your sample'" at a cost of a few thousand dollars, "people look at me like I'm crazy."
Chris Adams, who handles proteomics analyses at the Stanford University mass spectrometry center, said that any facility is "defined" by its available instrumentation, so it's important to communicate that to a client from the start.
"When clients come in with some problem, biological problems that they're interested in investigating, they might have some paper where someone did some targeted approach, I'll make it quite clear right away that this is not something that we can do right now," Adams said. "Setting the expectations that your collaborators have to a realistic goal is very important."
David Friedman, associate director of the proteomics lab in the Mass Spectrometry Research Center of the Vanderbilt-Ingram Cancer Center added that figuring out how to charge clients when the work entails the highest-end technology is especially challenging.
"How do you charge for what you really do?" he asked. Key to the process is communicating to the client "way ahead of time" before the work is started, exactly what will be done, what problems and bottlenecks may need to be overcome, and what kind of data may result, he said.
"At the end of the day, the reason we think we're effective at this is, win or lose, publish or not, data or not, they not only end up coming back for another experiment, they actually pay the bill, as well," he said.
Adams said that staff at a core facility should also utilize the brain- and manpower that the client offers, and as best that it can, a facility should check that adequate funding will be available to carry the project through to finish.
So what almost always guarantees that a project from a client will have serious problems or downright fail? Myers cited issues with samples, particularly if it comes from unfamiliar source material. Walk-in samples, he added, are especially problematic and have a "higher than normal rate in my lab."
If either he or the client is not engaged in the project and its results, it probably will be headed for failure, he said.