According to theory, comparing the expression levels of proteins in disease and healthy tissue using mass spectrometry should help researchers pick out proteins associated with a disease, and therefore likely drug candidates.
So much for theory.
In practice, teasing out the differentially expressed proteins is where one of the true challenges lies. But MDS Proteomics, which has bet heavily on differential proteomics as its strategy for drug discovery, may have found a shortcut. At the American Society for Mass Spectrometry meeting earlier this month in Orlando, Fl., MDS presented new software for automating the identification of differentially expressed peptides — shaving weeks off their analysis time, the company says.
The software, imaginatively termed DAS, for Differential Analysis Software, employs a series of proprietary spectral analysis algorithms to run through the output from MDSP’s Applied Biosystems/ MDS Sciex QSTAR Q-TOF and Fourier Transform Ion Cyclotron Resonance (FT-ICR) mass spectrometers, picking out the differentially expressed peptide peaks for subsequent MS/MS analysis. As a result, MDSP researchers told ProteoMonitor they have increased their throughput 50-fold.
“Before the software, you basically had to dig through the data manually, and find interesting ions and check the samples to see if they were in multiple samples, or if they had an isotopic pair,” said Jennifer Caldwell, a research scientist at the MDS Proteomics facility in Charlottesville, Va., who gave a talk on the software at the ASMS meeting. “It could take weeks to get a list of 20 or 30 good differences that you could then go in and do directed MS/MS on to get peptide and protein identification.
“The software provides us an easier way to get through that step of the data processing,” she added. “We can feed our mass spec data into the DAS program, and in the course of a day, easily get 100 good differences, if not more.”
Although attendees of the ASMS conference expressed interest in learning how MDSP constructed the DAS program, which was written by a team of programmers led by Chris Orsi at MDS Proteomics’ headquarters in Toronto, the company has no plans to make the software publicly available, Caldwell said.
While most academic labs attempt some form of automated matching, unlike the DAS program these approaches still involve a fair degree of manual analysis to achieve the same degree of confidence, she added.
MDSP primarily uses the software to analyze two types of experiments: to compare healthy and disease tissue or fluid samples as a means of identifying potential drug targets; and to track the effects of drugs in patients using samples taken at various stages of disease treatment, or compare treated with untreated samples, Caldwell said.
The DAS program is also compatible with experiments comparing both stable isotope-labeled peptides, and two or more unlabeled peptide samples, she added.
In her talk at the ASMS meeting, Caldwell gave several examples of how MDSP had employed the DAS program. In one, she explained how DAS had sifted through replicate analyses of a differential protein expression experiment on the company’s nanoliter HPLC, microliter electrospray FT/MS platform, located in Charlottesville, to analyze 200,000 signals from each run.
One of the most valuable aspects of the DAS program is the flexibility of its search parameters, added Jarrod Marto, a mass spectrometrist at the MDSP facility in Charlottesville. “Many times depending on experimental design you may have some a priori knowledge that the things you’re going after are relatively abundant,” Marto said. “But most of the time you don’t [and] there isn’t necessarily any correlation between absolute signal intensity and biological relevance. You have to be able to inventory and query everything, all they way down to low signal to noise values.”
Added Orsi: “For a person to do that, the time would just increase exponenetially.”