|
May 28 - June 1, 2006, ASMS Conference on Mass Spectrometry 2006, Seattle, WA, USA
Why is Automating the Determination of Molecular Ions Using Automated Approaches so Hard and How Might it be Used?
Mark A. Bayliss1; Vitaly Lashin2
1Advanced Chemistry Development Inc, Toronto, ON;
2Advanced Chemistry Development, Moscow, Russia
Abstract
Introduction:
Molecular ion identification is the corner stone of all MS1 data analyses that are undertaken and which forms a major time allocation when it comes to data reduction and primary assessment by all manner of scientists across all sectors of mass spectrometry disciplines. A number of highly accomplished algorithms have been created over the years using chemometric approaches applicable in the reduction or removal of low frequency noise. Such examples including CODA from Windig et al, MEND from Karger et al., however these algorithms can be classified as extremely sensitive peak extraction algorithms as the output requires a scientist to analyze the contributing ions as a particular rention time solving for molecular weight and key information.
Methods:
Using CODA as a specific peak extraction algorithm, a novel approach to retention time grouping was developed to overcome the limitations imposed by the inherent noisy nature of the MS1 data. To limit the amount of optimization that has to be carried on a per data set basis, a self-modeling approach to data filtration and selection had to be developed. To determine the molecular ion for each eluting component, it was necessary to determine all ions contributing to a single ion cluster within a particular retention time region. The determination of molecular weight was made possible using as much of the contributing ion clusters as possible, solving for classical adducts, multimers and ion losses from the molecular ion 12C.
Preliminary Results:
Solving for molecular ion, the goal of this project represents a major challenge both mathematically and from a mass spectrometry point of view. Often, even in electrospray the presence of radical cations and peak artifacts can cause miss-assignment of the 12C on which the determination of the molecular ion is key. The presence of competing ion clusters in the component spectrum in many cases may be as simple as a single molecular ion, however in more complex spectral scenarios, adducts and labile fragment ions challenged our ability to correctly deconvolve the spectral elements. A challenge which faces all scientists today is how to differentiate real peaks from noise and noise clusters that appear as peaks either naturally or because of the necessary processing actions such that it appears somewhat peak like. Our approach using multiple data elements to provide a component analysis has helped to overcome some of these challenging and what may turn out to be impossible challenges to overcome with total confidence. During our investigations, we have found that once all algorithms were brought together into a cohesive series of actions, that it is technically possible to apply this approach to a wide number of samples from a wide number of application areas. In one approach that has been investigated, the combination of LC/MS on a time of flight, operating in accurate mass mode, with source induced dissociation applied and with applied mass delta knowledge, fractional mass filters, and empirical formula generation as function of every extracted 12C ion in the dataset, we were able to demonstrate extensive coverage of chemical space for related chemical species. Coupling the accurate mass analysis with automated structure and fragment searching provides a rapid means of screening complex datasets for related and specific compounds of interest.
Download the poster in Adobe Acrobat format (131 Kb PDF file).
Relevant Product: ACD/IntelliXtract
Relevant Solutions: Metabolite ID, Impurities and degradants, Analytical laboratory
|
