An improved peak clustering algorithm for comprehensive two-dimensional liquid chromatography data analysis.

Affiliation

School of Food Science and Engineering, South China University of Technology, Guangzhou 510640, China; Guangdong Food Green Processing and Nutrition Regulation Technologies Research Center, Guangzhou 510640, China; Beijing Advanced Innovation Center for Food Nutrition and Human Health, Beijing Technology & Business University, Beijing 100048, China. Electronic address: [Email]

Abstract

In this work, an improved algorithm was developed for two-dimensional (2D) peak detection in complex two-dimensional liquid chromatography (LC×LC) data sets. In the first step, conventional one-dimensional peak detection was performed. In the second step, retention time, bidirectional overlap and unimodality criteria were applied to decide which of the individual peaks should be merged. To improve the peak detection with LC×LC analysis using shifting second dimension (2D) gradients, the variable thresholds, which permitted different thresholds for candidate peaks at different first dimension (1D) retention times, were employed for examination of the 2D retention time differences. Furthermore, the bidirectional overlap criterion performed at specified height was recommended to improve detection for tailing peaks. The developed algorithm was further tested on data sets from different LC×LC analyses of a complex peptide mixture, and then quantitatively evaluated by comparison between the results by the algorithm and mass analysis. Evidently improved performance with an accuracy rate over 60% was obtained by the algorithm, even for peak detection with LC×LC analysis under relatively low 1D sampling frequency or shifting 2D gradients. This would help to improve LC×LC quantitative analysis and performance assessment.

Keywords

Data analysis,Overlap criterion,Peak detection,Retention time criterion,Two-dimensional liquid chromatography,

OUR Recent Articles