NEW YORK (GenomeWeb) – Since Sciex launched its Swath data-independent acquisition mass spec method in 2011, DIA mass spec has carved out a significant space within proteomics.
Swath and Swath-style workflows developed by other vendors (Thermo Fisher Scientific, most notably), have seen rapid uptake among proteomics researchers drawn largely by the method's quantitative capabilities, which potentially offer higher throughput and higher reproducibility across samples than traditional data-dependent mass spec methods.
Some core labs, however, have remained hesitant to incorporate DIA mass spec as part of their fee-for-service offerings, with lab directors citing questions around method development, software capabilities, and customer benefit as factors that have slowed their adoption of the approach.
Traditionally, proteomics experiments have used data-dependent acquisition, wherein the mass spectrometer performs an initial scan of precursor ions entering the instrument and selects a sampling of those ions for fragmentation and generation of MS/MS spectra. Because instruments can't scan quickly enough to acquire all the precursors entering at a given moment, many ions – particularly low-abundance ions – are never selected for MS/MS fragmentation and so are not detected.
In DIA, on the other hand, the mass spec selects broad m/z windows and fragments all precursors in that window, allowing the machine to collect MS/MS spectra on all ions in a sample. This means that, unlike in DDA where values can be present for a protein in one sample and missing in another, DIA datasets are highly reproducible across many samples, which improves quantitation and the ability to, for instance, evaluate the levels of different biomarkers across a number of conditions or samples.
Yet, despite this potential advantage, many proteomics core labs are still uncertain as to whether DIA makes sense as a fee-for-service offering.
"It's a revolutionary technology," said Emily Chen, director of the shared proteomics resource at Columbia University Medical Center's Herbert Irving Comprehensive Cancer Center. However, she added, "the practicality [from a core lab perspective] hasn't been worked out yet."
DIA's potential for better, more reproducible quantitation "might be attractive," she said. "But we have to be able to present [to customers] how they benefit from using this new technology for their science, not just better quantitation. And we have to be able to clearly [establish] what we can deliver. If you can't say, 'I'm going to deliver X,Y,Z,' then you can't set the price for it."
Chen said that she has been using DIA mass spec in her own research for around three years, but the she has yet to implement it in her core lab. She is organizing a DIA workshop at next year's Association of Biomolecular Research Facilities annual meeting to bring together vendors and labs to discusses the challenges of implementing the approach.
Brett Phinney, manager of the proteomics core at the University of California, Davis, said that he sees great potential benefit from DIA's greater quantitative reproducibility.
"When you are trying to do label-free quantitation with DDA, there can be a lot of missing values," he said. "Sometimes, you can get up to 60 percent missing values. So, I think DIA has really great potential."
However, he added, he considers existing software and data visualizations tools inadequate and is hesitant to spend too much time working on new method development given the financial imperatives of running a core facility.
"My core facilities [run on] almost all soft money," Phinney said. "I have to support all of our work and all of my people through the money I charge [customers]. So, one of the problems I have is, there are so many different flavors of DIA and Swath, and I don't have time to test all of them. What size [m/z] windows do I use? Do I use small Windows? Do I use overlapping windows? I don't know."
Beyond questions of method development, Phinney said that options for visualizing DIA data are limited. He cited Skyline, a targeted proteomics software package developed by the lab of University of Washington researcher Michael MacCoss, saying that while it is a great program for relatively targeted analyses, it is not as well suited to looking at data from large-scale DIA runs, which can number in the thousands of proteins.
"That's a lot of data to look though," he said. "And, I haven't found the right way to visualize that data and return it to my customers in a way that I'm happy with quite yet."
To an extent, the desire to use DIA to look at such large-scale datasets is at the root of the problem, MacCoss suggested, noting that while much of the discussion around the technique has revolved around how many proteins it can identify and quantify compared to traditional DDA mass spec, DIA is, in fact, better suited to narrowly focused investigations.
He said that he and his colleagues are working to improve Skyline's tools for visualizing the large quantitative datasets DIA methods can generate, but that it is ideally an approach for looking at quantitative data on, for instance, tens of proteins, rather than thousands.
Where DIA could provide significant value for core labs is in its ability to collect quantitative data on large numbers of proteins than can be stored and analyzed in the future if desired.
Developing a targeted mass spec assay using conventional approaches like multiple-reaction monitoring "is extremely laborious," MacCoss said. "You have to pick the right peptides, pick the retention times, make standards if you want them."
"What you will often have is someone will come to you with a set of samples and say, look, I want you to measure the quantity of these 10 proteins," he said. "You spend a lot of time building an assay to those proteins, and you return them and they say, 'That is great. Now can you go measure these other 10 or these five new ones.' But then, of course, you need to make a new assay and redo it in new samples."
In theory, a DIA experiment would have collected quantitative data on all of these proteins, so instead of having to run a new experiment for the customer, a lab could reanalyze the already collected DIA data to extract the levels of these new proteins.
"DIA offers a really nice alternative because you don't need to do any upfront design of that assay," MacCoss said. "You don't need to pick which peptides you are going to use today and then be stuck with them five years from now, because you are measuring, in theory, anything that falls within a given mass range."
Such a more narrowly targeted approach is in line with what Chen said many core lab customers are looking for.
"Getting more [proteins] is really not the issue, because a lot of people are already overwhelmed with the number of markers they get," she said. "If you provide someone with 100 [candidate markers], how are they going to validate all of them? So, getting the number up is not attractive for the regular research laboratory."
In a way, though, this could reinforce core labs' decision to stick with DDA, she suggested. While quantitative data generated using DDA-based techniques like isobaric labeling might not be as reproducible across multiple samples as DIA assays, core labs simply discard from their analyses proteins with poor reproducibility, Chen said.
"We eliminate those, but we are still getting good data for [research customers] to move forward," she said. "So, what is the motivation?"
Another issue is the fact that to extract quantitative information from DIA data, researchers must first generate a spectral library for their sample, which requires an initial DDA experiment. This means more work for the core lab. A lab could, in theory, charge extra for DIA data based on the notion that it is higher quality quantitative data, but, Chen said, "I don't think anybody will go for it."
She noted, though, that researchers have been compiling spectral libraries for commonly used sample types, and she believes that in the next year or so there should be enough public libraries for humans, mice, certain cell lines, and plasma that researchers will not have to build spectral libraries of their own.
"If you work on anything else you'll still have to, though," she said.
In terms of vendor support for DIA, Chen noted that with respect to the Swath-style method first introduced commercially by Sciex and later implemented by Thermo Fisher on its Q Exactive and Orbitrap Fusion Lumos instruments, there are issues surrounding the underlying IP that has prevented Thermo Fisher from offering robust in-house solutions for dealing with DIA data.
Representatives of both Sciex and Thermo Fisher declined to comment on any legal questions around their DIA products, but it is the case that Sciex has developed its own DIA analysis tools while Thermo Fisher has looked to outside partners to develop DIA tools.
"Our approach, has been to partner with organizations that are both expert in the application of the DIA technique and, consequently, in the specific challenges of accurately processing DIA data," Ken Miller, Thermo Fisher's vice president of marketing, life sciences mass spectrometry, said in an email.
MacCoss is one of its main partners in this effort. The other is Biognosys, a Swiss proteomics firm focused on targeted proteomics and DIA analysis. Though it is a spinout from the lab of Swiss Federal Institute of Technology Zurich (ETH-Zurich) researcher Ruedi Aebersold, who developed the Swath approach commercialized by Sciex, the company is heavily focused on developing DIA analysis tools for Thermo Fisher instruments.
Phinney, who has used Thermo Fisher mass specs for his explorations of DIA, said that he had used Biognosys' DIA software package, Spectronaut, and thought it looked promising. However, he said he was not able to test it as thoroughly as he would like under the terms of the trial provided by Biognosys.
"I wish I would have had more time to play around with it and form an opinion on it," he said, noting that, while it is an expensive piece of software, with annual single system licenses running $14,000, he's not inherently opposed to spending the money.
"I don't mind paying for software, right? But that's a big investment, and I want to make sure it works for me and my clients before I purchase something," he said.
Chen said that she and her other ABRF colleagues are negotiating with Biognosys on behalf of the organization about potential deals for access to the Spectronaut software.
In a study published last year in Nature Biotechnology, researchers at the University of Mainz in Germany benchmarked the performance of five commonly used DIA mass spec software programs including Spectronaut, Sciex's Swath 2.0, and the open-access packages OpenSwath, Skyline, and DIA-Umpire. They found that the packages, excluding DIA-Umpire, were essentially equivalent in terms of performance, and that DIA-Umpire, while at the moment behind the others, showed great potential to improve as mass spec instrumentation grows more powerful.
However, the study did not focus on items like the ease of data visualization or delivering results to customers, which, as Phinney noted, are key concerns for core labs. It also looked only at the performance of the software in analyzing data generated on Sciex instruments.
Mark Cafazzo, director of Sciex's academic/omics business said that the advantages of DIA are perhaps most notable in studies where researchers are running hundreds or thousands of samples, which does not necessarily fit the profile of some core labs' customer bases.
For core labs doing mostly smaller one-off experiments for research customers, DDA approaches like isobaric labeling might be a better fit, he noted.
That said, "We definitely have had feedback from certain core labs where they have come to us looking for strategies and concepts they can use to promote Swath to their customers," he said.
Christie Hunter, director of omics applications at Sciex, noted that the company does work with a number of researchers "who are working in highly collaborative environments where the majority of their work is running sample sets for their collaborators, and a lot of them are using Swath."
She added that Sciex considered such labs in the design of its OneOmics cloud project, a collaboration with Illumina to put omics analysis tools in a cloud environment.
"Part of the thinking going into the OneOmics project was that cores could generate the data and get the results up into these nice viewer applications that they could share with their collaborators," she said. "And their collaborators could then look at [the data] in the biological context, so they didn't feel like they were looking at mass spec data but looking at biological data."
Phinney said that he plans to continue work on implementing DIA in his facility, despite what he sees as the challenges involved. He noted that ABRF's proteomics research group is planning a study on DIA in core labs that could help facilities incorporate the technique as part of their offerings.
"Hopefully, we'll be able to send the sample out so people can run DIA on it, and we'll get some kind of feedback on what works well for core facilities," he said. "I think DIA has a lot of promise."