Increasing the Number of Analyzable Peaks in Comprehensive Two-Dimensional Separations through Chemometrics

Abstract

Comprehensive two-dimensional (2-D) separations are emerging as powerful tools for the analysis of complex samples. The substantially larger peak capacity for a given length of time relative to 1-D separations is a well-known benefit of comprehensive 2-D separation methods. Unfortunately, with complex samples, the probability of peak overlap in 2-D separations is still quite high. This is especially true if one desires to speed up the analysis by reducing the run time and, thus, by reducing the resolving power along the first dimension separation. Chemometric methods hold considerable promise to overcome the limitations brought upon by the likelihood of peak overlap. Thus, chemometric methods should be able to effectively extend the resolving power of 2-D separation methods. In this paper, the theoretical enhancement provided by application of the generalized rank annihilation method (GRAM) for the analysis of unresolved peaks in comprehensive 2-D separations is carefully modeled and critically evaluated. First, Monte Carlo simulations are used to determine the conditions where the use of GRAM results in the successful analysis of unresolved peaks. A wide range of experimental conditions and performance criteria are modeled, typical to many available 2-D separation methods, including analyte/interference peak height ratio, first- and second-dimension resolutions, signal-to noise ratio, injection volume reproducibility, and run-to-run retention time reproducibility. Essentially, a wide range of experimental conditions and performance criteria are found to provide reliable data amenable to GRAM analysis. The information gleaned from this first set of simulations is then used in conjunction with Monte Carlo simulations of comprehensive 2-D separations. For these simulated 2-D separations, the total number of analyzable peaks when using GRAM was determined and found to be substantially better than using only traditional quantitative methods such as peak integration or height. For example, it was determined that the use of GRAM increases the average number of analyzable peaks by a factor of 2 for 2-D separations in which the peak capacity is 67% occupied by randomly distributed peaks. The results of the studies are general, and the use of GRAM should increase the number of analyzable peaks for all forms of comprehensive 2-D separations.

Comparative analysis of peak-detection techniques for comprehensive two-dimensional chromatography

“…Similarly, the issue of coelutions, although important for peak detection, was not considered herein, as neither the two-step nor watershed algorithm incorporates a solution for coelution. Other techniques have been developed for unmixing coeluted peaks, e.g., [11][12][13][14][15][16][17][18][19][20][21]. Step (2-Step) peak detection algorithms for peaks with parametric skew (s); first-column (x-dimension) peak-width standard deviation, σ x ; second-column (y-dimension) peak-width standard deviation, σ y ; and noise standard deviation, σ n .…”

Processing and classification of protein mass spectra

“…In the absence of noise the number of linearly independent components or the PROCESSING AND CLASSIFICATION OF PROTEIN MASS SPECTRA & rank of the data matrix S(t,m/z) equals the number of signals (Fraga, Bruckner, & Synovec, 2001). Generalized rank estimation methods, which provide a robust estimate of the number of independent components, therefore provide an estimate of the number of signals.…”

Comprehensive multi‐dimensional liquid chromatographic separation in biomedical and pharmaceutical analysis: a review

“…Standards with their precise retention time are required and there must be some resolution in both dimensions (Fraga et al, 2001). In one approach (not MDLC), a pair of stationary phases with different selectivities was run in parallel, with a split sample injection.…”