Preprocessing of hyphenated analytical data

In the field of life sciences, LC-MS and GC-MS are among the most popular platforms to acquire data from human, microbial and plant tissues for food, nutrition and health applications. From these raw data, the metabolites can be derived that are most important and influential for a certain application. However, besides a proper data quality and analysis strategy, this also requires a preprocessing step that allows extracting information of pure components from the data. Deconvolution is an approach to perform such tasks.

The starting point of this project is an existing tool for the deconvolution of GC-MS data (i.e. TNO-DECO, see Jellema et al, 2010. Chemometrics and intelligent laboratory systems 104: 132). Its unique building blocks are used to build a generic and automatic pre-processing tool for high resolution (HR) LC-MS and GC-MS data.

More technically, alternatives for the engine of deconvolution (MCR-ALS: Multivariate Curve Resolution – Alternative Least Squares) have been developed using technologies that are used in the field of image processing. Furthermore, a binning approach has been developed that reduces the variability in the HR accurate mass numbers leading to lower computational effort and more reliable deconvolution results. Moreover, a baseline correction and noise removal approach has been developed that seems to be essential for deconvolution of LC-MS data. As the results of these new approaches are very promising, several publications are in preparation. In the near future, development will be focussed on using prior knowledge and figures of merit. This will further increase quality, objectivity and automation of the procedure. 

Main project title: 
Data pre-processing
PD 1-4-09
Principal Investigator: 
Code 1: 
Code 2: 
Status Project Proposal: 
Location 1: