WITHDRAWN: Clustering of Tandem Mass Spectrometry Data

• Author(s): Xiaoyu Yang and Pedatsur Neta and Stephen Stein
• Keywords: clustering ; dot product ; mass spectrum ; spectrum similarity ; binning

Abstract

Clustering techniques have been applied to mass spectral data analysis. A distance-based clustering algorithm is used for mass spectrum grouping or clustering. For each pair of MS/MS spectra with equal precursor m/z, dot product is computed as a measure of spectrum similarity. The spectrum with the highest average dot product is used as a cluster center. If the dot product value is equal or bigger than 0.9, then the two spectra are put into the same cluster. The transitive closure rule is also applied in the clustering. Within each cluster, binning-based peak clustering algorithm is used to generate the consensus spectrum. The algorithm is tested using spectra from various mass spectrometers and from different samples. The consensus spectra are inspected compared with raw data. This algorithm is used in building the NIST Mass Spectral libraries and quality control.