Know more

As high-res ultra-sensitive analytical systems coupling gas-phase chromatography with mass spectrometry (GC–MS) gain increasing currency, volatolomic signatures (i.e. the volatile fraction of a biological sample) are fast becoming increasingly complex and datadense. Sifting through these volatile organic compound signatures to extract the discriminant information thus requires powerful and purpose-geared chemometrics. Here we set out to compare performances across three exploratory unsupervised multivariate methods to tease out markers of exposure from the volatolomic signatures.
Working up from a dataset of liver-sample volatolomes from animals exposed or not to chemical contaminants, we compared three chemometric approaches: principal component analysis (PCA) on ANOVA-prefiltered data, independent component analysis (ICA) and common component analysis (CCA). This study underscores the value of common component analysis to (i) discriminate different subpopulations exposed or not to different key families of foodborne micropollutants, and (ii) uncover candidate markers of chemical-risk exposures.

Before adopting CCA as a benchmark method for screening volatolomics datasets and ultimately applying it to other omics datasets, we are first doing further research to verify that candidate markers uncovered through this study fit with markers found in independent experiments.