“Once we’ve done this initial stage of improving the quality of the data coming in, it opens the floodgates for using a lot more powerful statistical techniques and bringing things in like multivariate analysis.”
Nonlinear Expands Progenesis Line Beyond 2D Gels to Tackle LC/MS, Multivariate Stats
Nonlinear Dynamics has built its business around 1D and 2D gel analysis, but this week the company took steps to expand its flagship Progenesis line of 2D gel-analysis software into different segments of the proteomics analysis pipeline: LC-MS and multivariate statistics of data from multiple analytical platforms.
The foundation for the two new products, Progenesis LC-MS and Progenesis Stats, is an image-analysis algorithm called SameSpots that the company launched last year. Nonlinear claims that the SameSpots technology is able to extract more information from 2D gels via improved image-alignment methods as well as background subtraction and normalization techniques.
Now, Nonlinear said it has applied that same approach to LC-MS data, which in turn enabled the company to offer improved statistical tools for analyzing both 2D gel and LC-MS data together in the same context.
The UK-based company hopes that Progenesis LC-MS will expand its footprint in the US market, where proteomics researchers favor liquid chromatography over 2D gels, and that Progenesis Stats will appeal to labs running 2D gel and LC-MS applications.
“Once we’ve done this initial stage of improving the quality of the data coming in, it opens the floodgates for using a lot more powerful statistical techniques and bringing things in like multivariate analysis,” David Bramwell, technical director of Nonlinear Dynamics, told BioInform’s sister publication ProteoMonitor this week at the Association for Biomedical Research Facilities conference in Tampa, Fla.
“We found that the major problem with 2D gel analysis — and to a degree, most of the LC/MS analysis as well — is the alignment problem,” Bramwell said. This drawback has caused proteomics to fall behind other -omics disciplines like microarray analysis in terms of statistical rigor, he said. “Microarrays are actually doing a lot more robust statistical analysis, mainly because all the data lines up.”
In addition, he noted, proteomics has been hampered statistically by huge amounts of missing data.
“If you went up to most research statisticians and showed them the amount of missing data in proteomics data, they would just walk away and say, ‘Come back when you’ve got a decent data set,’” he said.
“Now that we’re actually extracting the information correctly, you don’t really need a lot more advanced statistics,” he said. “The statistics work better. The statistics that people have been using for years like t-tests and ANOVA are robust.”
Bramwell noted that one key advantage of overcoming the missing values challenge is that researchers can further improve the statistical significance of their experiments by running more replicates.
With traditional analysis methods that are unable to extract all the information from an experiment, researchers often ran into a “paradox,” he said. “As you tried to go for statistical robustness by having more replicates, your data kind of got noisier and smaller.”
Now, he said, Nonlinear is working with pharmaceutical companies on experiments with up to 60 replicates. The company had a poster at ABRF in which it detailed an experiment with 18 sample groups of 20 replicates per group, “which was an enormous undertaking previously,” he said. “Just getting that data into a form that you could start doing the analysis on … would have taken months” without the SameSpots software, he said.
Bramwell noted that the launch of Progenesis LC-MS should help the UK-based company expand its presence in the US market, where proteomics researchers overwhelmingly favor liquid chromatography over 2D gels.
He also noted that the Progenesis Stats product, which enables researchers to analyze 2D gel and LC-MS data together in a single statistical environment, should appeal to labs running both platforms.
In addition, he said, there’s a possibility that the SameSpots technology could help drive a resurgence in adoption of 2D gel analysis. “We think that a lot of people moved away from 2D gels thinking they weren’t really reproducible, but they like the coverage and how they’re really good as a discovery tool,” he said. “Now … when we talk to people about it, quite a few people are considering it again because it’s a really powerful technique,” he said.
— Tony Fong, editor of ProteoMonitor, contributed to this article.