Making a large purchase for your lab can be a daunting and time-consuming process. There's a lot of information to sort through to be sure that you are getting the best instrument for your needs.
Everyone has a slightly different strategy for evaluating the tools on the market. Columbia University's Lewis Black gathers as many opinions as he can while at meetings or visiting other institutions, while Dick Bennett at Children's Hospital Boston makes spreadsheets to evaluate different aspects of the instruments. Many researchers insist on trying machines out before committing their funds. But they are all curious about the same thing: does the machine have the specifications they require?
"You really want a stable platform, so you want to make sure the platform's stable before you buy it," says Vince Magrini from Washington University in St. Louis.
Among the issues you should consider are whether you really need a new instrument, whether you have the capability to store the data coming off of it, and whether you have the funds not only to purchase it, but also to pay for its related service contract.
Do you need it?
Buying a new instrument can be broken down into two categories: purchasing a completely new piece of technology for the lab and replacing an old piece of equipment.
Before committing to buying the latest shiny toy, you have to be sure it is worth the expense — that buying this new item will help your research or, if you are a core lab, it meets the needs of your users. "You've got tons of salespeople trying to sell you the next thing from every company in the world," says Lisa White, who directs the microarray core facility at Baylor College of Medicine. "The problem with that is that you have to show that you are actually going to have business if you are going to try to purchase it." To determine which instruments to bring into her core lab, White surveys center directors and department heads by e-mail to gauge their interest in a new technology.
The plans for a machine should also expand beyond the immediate future, adds Gregory Shipley, director of the Quantitative Genomics Core Laboratory at the University of Texas Health Science Center at Houston. "You are really going to have to sit down and think about not only what you might want to do with [a new machine] today, but what you might ever want to do with it in, I would say, the next five years," he says. "And be very candid with yourself. If you think you might want to have capability to do some more fancy things than you are doing right now, then you better build it in now."
It's also helpful to take a look at the data that comes off of the machine you are considering, to see if it really provides information that will be useful to you. "I look at the data first and then the software that produced that data. I think to myself: 'Is that the kind of results that I want? Are these results structured in a way that will be useful for me and my collaborators?' And if they are not, then that's a big minus," says Lewis Brown, who runs Columbia University's Comparative Proteomics Center.
Replacing a worn-out instrument also requires consideration, as funds can be difficult to come by, Baylor's White adds. "Everybody's all gung-ho about getting new technology," but when it comes to "replacing old technology and stuff they are sort of like, 'Eh,'" she says. "It's not exciting, but if it's a workhorse, then definitely, it needs to be done."
However, many of the same considerations — unless you want a straight exchange of equipment — that need to be thought about for a new instrument also need to be considered when buying a replacement instrument.
To know what the options are for a new device and to help figure out which ones fit best in certain workflows and meet particular needs, many researchers turn to the literature, meetings, and networking with colleagues — or a combination of all three — to keep abreast of the latest tools.
Columbia's Brown relies extensively on meetings and his network of colleagues to figure out which proteomics tools suit his lab. For his most recent purchase, he began by attending sessions and speaking with poster presenters and vendors at an Association of Biomolecular Resource Facilities meeting. "I think about it in my own mind as 'interrogated the community,'" he says. The vendors, he adds, sometimes tell him when certain posters show their products performing certain applications and giving certain data. In this case, Brown tracked down the posters to see the data and would "really interrogate the author of that paper to see if that was really true," he says. And though he settled on one particular vendor from that exercise, he still repeated the process at the American Society for Mass Spectrometry meeting. After that meeting, Brown spoke with people at other institutions about similar equipment and asked them about their experiences with it. "It's a challenging and exhausting process which requires a great deal of networking, understanding of the scientific literature, the ability to sift between commercial claims and scientific performance," he says.
To keep all of that information straight, Children's Hospital Boston's Bennett creates a spreadsheet to rank the different specifications that he wants so that he can compare different vendors on the same metrics.
In addition to attending meetings herself, White says her users often come to her with ideas that they have picked up from the literature or meetings about what tools her core should consider purchasing. Then, part of her due diligence into a technology includes reading the literature herself and speaking to other core lab directors about how they support their machines. Sometimes she is able to bring that new technology in, though other times she cannot or she directs a user to another core that is better suited to handle that specific tool.
Alexandre Montpetit, assistant scientific director at the McGill University and Génome Québec Innovation Centre, also keeps an eye on what the larger sequencing centers are doing as a way to gauge the direction of the field. "You might not want to be in a position where you are going with a technology that might be good, but nobody else is using it and in the end you are going to be stuck with that equipment," he says.
Having that established user community is also important to Brown. One of the questions he asks himself about a new technology's user base is: "Is it robust enough so that I am not the only person out there?" While he notes that some researchers are technophiles who have the resources to build up an instrument's technology and software, he is not one of them. A good base, he adds, can mean that there are enough users in the area to support good service.
Who you know
Sometimes staying with a favorite vendor or two whose instruments you know well can be helpful. After all, if you have already worked with that company, you know their sales and service representatives, and that their tools fit into your workflow.
However, it's not always practical to do so, so be careful not to get too attached. "I think in my case there was the reliance on my experience with instrumentation that I used elsewhere, and then it becomes the devil I know versus the devil I don't know," Columbia's Black says. "I know all of the little deficiencies that that vendor has — the dirty little secrets."
He adds that if he is buying an instrument that is similar to what he already has and is not looking for different capabilities, he is likely to stick with the vendor he already knows. "It would not resonate with me, the small improvements that another instrument would provide because one would then have to re-learn how to use that instrument for very little benefit," he says.
Indeed, the change might not be worth the time it would take to figure out how to make the new system run well if there is not much difference between them. "You have a machine from company A, even though you've heard really good things about company B — the software, for example. Going from Applied Biosystems to the Roche software is like Venus and Mars. It really is a whole different way — and they both work well — but it's just a matter of figuring out how to make the other one work," UT's Shipley says. "People will often buy from the same company because they know they can deal with the software and they don't have to go through that hurdle." His core made that argument back in the 1990s when he was purchasing a second 770, he adds.
Montpetit says that if you are just expanding your lab's number of machines, then it is easier to go with what you already have. For example, you already have your protocols and bioinformatics pipelines in place — why reinvent the wheel? If each machine has its own system, it can be a "disaster," he says. "It's much easier to stay, if you can, with the same technology, to have everything the same level. If you upgrade, upgrade everything."
However, the choice should be driven by which tool is the best for you and your needs. "With respect to the technologies themselves, just because we've worked with them in the past [and] we like their previous technologies doesn't mean that their newer technologies actually are going to be amenable to our needs," WashU's Magrini says.
When differences between technologies are profound, take the time to consider all the options. For example, Montpetit says that with third-generation sequencers, what machine you choose depends on what you need it to do — sequence single molecules, distinguish haplotypes, or determine base modifications. "That's where there might be a niche for different technologies," Montpetit says.
And this is not just true for sequencing. "When there is new technology out there, I think, at least in proteomics — where it's so incredibly competitive and fast-moving — that would be an unwise strategy to try to just have this brand loyalty," Black adds.
In addition, if one company is clearly the market leader, says Karen Jonscher, director of the Systems Biology Core Facility and proteomics center at the University of Colorado, Denver, she would be more tempted to go with its tools. "[If] they do the best job at it — that's something that weighs pretty strongly with me," she says.
Try it out
Another way to figure out what machine would be the best fit for your lab's needs is to try one as a beta-tester, use it for a month, or send some of your samples to the company to run on the machine. "Use the dang thing. See if it works for you, how it works with your workflow, and how it is to use because that's really the bottom line," Shipley says.
There are different kinds of centers or researchers, those that live at the bleeding edge of technology and those that are a bit further back from that edge. In sequencing, for example, new instruments are often beta-tested at large centers like the Sanger Institute, the Broad Institute, or WashU. "I'm in the unique position where we'll beta-test these things with the companies and get a feel for it," says WashU's Magrini.
Québec's Montpetit says that for large instrumentation, he keeps tabs on what larger sequencing centers are up to, and his group will often beta-test equipment smaller than sequencers. "Other centers, maybe of our size, would actually wait a little bit … get some feedback, know that it's been implemented there and it works and it does what we want. We do more evaluation of these very big equipments this way," he says, adding that at centers the size of his own, "usually it's not possible to get the machine evaluated. Smaller equipment, yes, but larger, no."
Baylor's White has also been an early-access customer, and says that having a piece of equipement in the lab can help her choose whether or not she wants to purchase that machine. "You'll get some equipment in and you're trying to run it and you find out there's all these bugs, or you see that something you thought might be helpful to the population is not and some other technology is easier or cheaper, or there's some point that is going to make people use it over this other one," White says.
If beta-testing is not an option, you can also try to get a machine from the vendor in the lab for a test drive. "I think that the best thing that — once you whittle it down to the top two or three that you might want — is to call the companies up and see if they'll bring one in and let you use it for a week or something," Shipley says. Bennett at Children's Hospital Boston says it is important to try a machine out in-house. He suggests getting a 30-day trial.
Alternatively, you could send your samples in to be run by the vendor on the machine in question, Bennett adds. That way you can see what kind of data it provides and whether it lives up to your expectations. Denver's Jonscher does that as well. "Send demo samples," she advises. "If you can, go to the demo yourself."
A growing concern with high-throughput technologies, particularly sequencers, is storing all the data that those tools generate.
"If they have their own building, that's a good thing," Magrini says. Indeed, WashU's Genome Center has its own 16,000-square-foot data center, chock full of servers.
Those without such luxuries have to consider the logistics of storing the data from any machine they are looking at, as storage costs can be high.
"That's a big concern. Especially when you go from a few [pieces of] equipment to a lot," says Québec's Montpetit. "There are things that you just cannot do, actually." His center recently hired a consultant to evaluate how it stores, handles, and transfers data from one server to another. Based on that, he says, the center made some changes. "Basically we cannot afford to keep the data, back up the data," he adds. Currently, his center keeps the intensity images for a week or two. While Montpetit says they had originally planned to keep the traces indefinitely, they are now weighing how long to save them. "Just with the amount of data that's generated, that's a few hundred gigs of just reads per run," he says. "We're debating right now whether it's six months or nine months or 12 months, but it won't be more than that for sure. Otherwise we'll have to spend a fortune on tapes to back everything up."
While data storage isn't yet an issue for Magrini, he says that cloud computing seems like a viable option for data storage. "But that's something that you need to consider, especially in your budgets — data storage, data processing, data transfer," he says.
Many machines are not sold, or run, all by their lonesome. Many come with computers from the vendors, preloaded with software, to run the instrument. As convenient as this sounds, UT's Shipley says it can sometimes cause problems. His institution does not allow anyone to purchase laptops for a lab because they are easily stolen. He currently has a ViiA 7 on trial from ABI, which came with a laptop. ABI provided a lock for it and accepted the risk that it may wander off. "If we had purchased that instrument, we would not be able to accept the laptop; we'd have to work something else out," he says.
In addition, Shipley's institution is loath to approve the purchase of a desktop from a life sciences vendor when it knows that it has a contract with a computer vendor and can get a similar computer for less. "Even though their software is loaded on there and all of that, it's sometimes hard to justify those purchases when the university knows it can buy the computer itself for a heck of a lot less money," he says.
He adds that vendors do work with him on that point and provide the software to be loaded on a computer purchased through the university. Other vendors, he says, can be more adamant. "Sometimes you are stuck. Just because they know that that particular model of computers works with their software and hardware, it's a hardware issue, actually," Shipley says.
When it breaks
To many people who deal with a lot of equipment, service contracts are indispensible. Many instruments come backed by a warranty, or a short period of time during which the vendor will help fix or troubleshoot problems. But after that time expires, a service contract — though pricey — can come in handy when things break. "You have to have it, you have to have it. That's a no-brainer," Magrini says.
Even the best machines go down. If you do not want to pay for the parts and labor out of pocket, service contracts will cover those expenses. In addition, a good service contract will include preventive maintenance. "They'll come by and they'll check and make sure the instrument is working properly," Shipley says, adding that in his case, for a real-time PCR machine, "they'll make sure that you don't have fluorescent contamination on the block or somewhere else on the head."
Denver's Jonscher says you should be sure that the contract does not include a maximum dollar amount or maximum number of visits for the same issue. The service engineers should "keep coming until it's fixed," she says.
Shipley also considers how fast the turnaround time on fixing an instrument will be. As a core facility, he says he cannot afford one of his instruments to be down for too long. "You want to know: 'Where would that guy come from? What are the odds it's going to get fixed within a week or something?' It'd be nice if it got turned around in a couple of days, actually," he says. While most of the companies he deals with have a representative in Houston, where he is located, someone in a smaller city may want to consider which firm is nearby.
One of the questions you'll have to ask yourself, Shipley says, is: "'How much is the annual service contract going to cost me?' Because believe me, after the first year, you have to think about that. 'Do I want to pay that? If I don't pay that, how much is it going to cost if something serious breaks?' If that guys walks through the door and spends an hour, you've almost thrown away $1,000 right there."
These contracts can be pricey, though — they can run from 5 to 10 percent of the cost of the machine. Finding funds to cover them can be difficult as well, Columbia's Brown says. "Service contracts are a big challenge for us in the academic sector because, particularly on large instruments, they are invariably going to be in some kind of shared or collaborative lab and the budgets for such labs are chronically low," he says. "The cost has been a really important factor in our thoughts and it really has inhibited us from buying instruments."
There is, though, some room for negotiation, Québec's Montpetit says. "It's easier if you have a lot of equipment from the same company. It makes a lot of sense, [and] you can negotiate that," he says.
With older pieces of equipment, it may no longer be prudent to keep a service contract, especially as it tends to eat into a budget. "You talk to the company: 'How much is it going to cost if you come here to fix it? If the laser breaks, what is it going to cost me?' Then you juggle and weigh those different costs and make a decision on it," Baylor's White says.
Unfortunately, she adds, you can end up in trouble either way. For one older machine, she decided to forgo the service contract. "The next week, the laser totally burned out and I had to spend quite a bit to get it fixed," she says.
Other times, the equipment is retired as it becomes more expensive to maintain. Montpetit notes that some small centers stop using machines once they can no longer afford the service contract. "I still have a lot of equipment in my room that is in the knee holes under the benches because I just closed it out," White adds.
Denver's Jonscher forgoes service contracts completely because of the cost — she maintains and fixes her proteomics center's machines herself. However, she says, service contracts are "crucial," and she would prefer to have them.
All in all, keep in mind what you and your lab need for the next five years or so. Shipley also advises caution about being sold something you do not really need. "Salespeople are really good about dazzling you with the technology," he says. "It's like buying a car, right, if you just put this package on with the racing stripe and, 'C'mon, you've got to have that.' No, you don't really. The car will still go down the road without the racing stripe."