A group of researchers from Katholieke Universiteit Leuven have demonstrated the application of principal component analysis (PCA) to imaging mass spec data.
At the 2007 Pacific Symposium on Biocomputing (January 3-7, Maui), the group presented the paper "Prospective Exploration of Biochemical Tissue Composition via Imaging Mass Spectrometry Guided by Principal Component Analysis." The authors note in the paper that there are typically thousands of ion images collected to compose each complete imaging mass spectrometry image data set, and that the volume of data can complicate unassisted scouting projects where the object is to find meaningful variations without the benefit of a priori knowledge of the identity or m/z values of relevant analytes.
"This is why we employ multivariate data analysis methods, such as the principal component analysis discussed in this paper, to perform a preliminary exploration of the data tensor in order to identify spatial and mass trends that merit further investigation."
In a prospective evaluation on rat spinal cord imaging mass spectrometry data, PCA found markers that differentiate white matter tissue from gray matter tissue, and dorsal tissue from ventral tissue, based differences in cellular protein composition between these tissue types.
"In summary, the PCA-results tell us that in this particular IMS data set the chemical composition is dominated by the difference between grey matter nerve tissue and white matter, and two quantitative ion markers for these areas are observed at m/z 5484 and 8564. In addition to that, a ventral/dorsal difference was measured which can be related to known ventral/dorsal differences in the spinal cord."
Raf Van de Plas, Fabian Ojeda, Maarten Dewil, Ludo Van Den Bosch, Bart De Moor, and Etienne Waelkens. "Prospective Exploration of Biochemical Tissue Composition via Imaging Mass Spectrometry Guided by Principal Component Analysis." (944 KB pdf)
Comments