Proteomics
Twin Peaks: Identifying peptides by direct comparison of experimental spectra
Comparison of peptide spectra is one of the most essential aspects of high-throughput proteomics, but it can be a bottleneck for laboratories without access to unlimited computing power. Like the murder mystery in David Lynch's epic TV series Twin Peaks, spectral comparison is really about solving a mystery of identity (in this case, the identity of a protein) in the shortest amount of time. So, what can be done to solve such real-life proteomics mysteries more efficiently?
The most widely used strategy for high-throughput protein identification is tandem mass spectrometry (MS/MS) coupled with high performance liquid chromatography (HPLC), followed by the comparison of peptide spectra with theoretical spectra predicted by protein sequences. However, one problem with the comparison of experimental spectra with those in databases is that they simply do not always match when they should. A recent article in Proteome Science indicates the possibility of an alternative approach - namely, the direct, pairwise comparison of experimentally derived spectra for the rapid identification of protein species.
In a collaboration between the McGill University in Montreal and the University of Toronto, a group of proteomics investigators led by Robert Kearney explored the possibility of identifying peptides based on the direct comparison of tandem mass spectra. Kearney and his colleagues focus on various computational aspects of this approach, demonstrating how to normalise peak intensities and how to vectorise spectra and measure their pairwise similarity using correlation coefficients. They also determined that ensemble averaging was a more effective technique for generating reference spectra.
Most importantly, however, this pairwise comparison method was computationally much faster than traditional database searches - a critical advantage as proteomics increases its demands for computing power.
Click here for useful links
