Global analysis of multiple gas chromatography–mass spectrometry (GC/MS) data sets: A method for resolution of co-eluting components with comparison to MCR-ALS

In most cases authors are permitted to post their version of the article (e.g. in Word or Tex form) to their personal website or institutional repository. Authors requiring further information regarding Elsevier's archiving and manuscript policies are encouraged to visit: a b s t r a c t a r t i c l e i n f o Global analysis has been applied to resolve components in multiple gas chromatography–mass spectrometry (GC/MS) data sets. Global analysis methodology is based upon a parametrized model of the observed data, including random (and possibly also systematic) errors. Each elution profile is described as a function of a small number of parameters. We successfully based the description of elution profiles on an exponentially modified Gaussian. The mass spectra were described non-parametrically. Model usefulness is judged by the quality of the fit and whether the estimated parameters that describe the elution profiles and mass spectra of components are physically interpretable. Advantages of the method are most evident with multiple data sets and overlapping elution profiles. Differences between data sets are described by alignment parameters and by relative amplitude parameters. The estimated mass spectrum is identical between experiments. Global analysis and multivariate curve resolution alternating least squares (MCR-ALS) are the only methods currently developed for component resolution for the case of completely co-eluting compounds in mass spectrometry data. In the present contribution global analysis is shown to have better performance than MCR-ALS in terms of the estimated mass spectra for a variety of simulated GC mass spectrometry datasets representing components that are completely co-eluting. The component resolution problem for an m × n matrix of data D 1 can be stated as the problem of estimating the matrices C 1 and S from D 1 in D 1 = C 1 S T ð1Þ such that the matrix C 1 is m × n comp , and each column represents the evolution of a component of D 1 in the variable with which the rows of D 1 are resolved, and such that the matrix S is n × n comp , and each column represents the component of D 1 in the variable with which the columns of D 1 are resolved. For chemistry applications the components often correspond to the different chemical compounds in the sample underlying the measurements, and the problem is sometimes referred to as deconvolution. Usually it is assumed that …