Abstract by Shiladitya Chatterjee
Chemistry and Biochemistry
PATTERN RECOGNITION ENTROPY (PRE) AS A SELECTION TOOL FOR MASS CHROMATOGRAMS IN LIQUID CHROMATOGRAPHY-MASS SPECTROMETRY (LC-MS)
Pattern Recognition Entropy (PRE) is a new take at Shannon’s statistical concept of entropy and information theory. Liquid Chromatography-Mass Spectrometry (LC-MS) datasets have noise arising from various sources such as electrospray ionization and mobile phase composition. As a result, noisy Total Ion Chromatograms (TICs) are often obtained. Here, PRE is used as both a selection tool and a summary statistic that is able to identify information-containing mass chromatograms (MCs), where chromatograms that are devoid of noise (richer in signal) have lower PRE values. PRE analysis was performed on an LC-MS separation of a proprietary surfactant mixture of at least 15 components. PRE values of the resulting MCs were plotted for each of the 1451 MCs. These values were then fit to a spline and a threshold value was set. The distribution of the differences between individual PRE values and the spline fit were used as the metric for selection of the mass chromatograms.The resulting reduction in mass chromatograms was significant (by an order of magnitude) and the TIC was significantly improved. The results were compared to a widely acknowledged algorithm (CODA).