Journal of Chromatography B (v.910, #C)

Chemometrics in chromatography by Pedro Araujo; Bjørn Grung (1).

► Design of experiments can efficiently optimise a chromatographic response by finding best values of factors. ► DoE with a Plackett–Burman design can be used in method validation to show robustness of a method. ► Central composite designs (CCD) are the most used for optimisation of 2–5 factors. ► Box–Behnken and Doehlert designs can be more efficient than CCD. ► D-optimal designs are used when the factor space is restricted or numbers of possible experiments is limited.The ability of a chromatographic method to successful separate, identify and quantitate species is determined by many factors, many of which are in the control of the experimenter. When attempting to discover the important factors and then optimise a response by tuning these factors, experimental design (design of experiments, DoE) gives a powerful suite of statistical methodology. Advantages include modelling by empirical functions, not requiring detailed knowledge of the underlying physico-chemical properties of the system, a defined number of experiments to be performed, and available software to accomplish the task. Two uses of DoE in chromatography are for showing lack of significant effects in robustness studies for method validation, and for identifying significant factors and then optimising a response with respect to them in method development. Plackett–Burman designs are widely used in validation studies, and fractional factorial designs and their extensions such as central composite designs are the most popular optimisers. Box–Behnken and Doehlert designs are becoming more used as efficient alternatives. If it is not possible to practically realise values of the factors required by experimental designs, or if there is a constraint on the total number of experiments that can be done, then D-optimal designs can be very powerful. Examples of the use of DoE in chromatography are reviewed. Recommendations are given on how to report DoE studies in the literature.
Keywords: Experimental design; Design of experiments (DoE); Response surface methodology; Method validation; Optimisation; Plackett–Burman design; Central composite design; Box–Behnken design; Mixture design; Doehlert design; Factorial design; Fractional factorial design;

Doehlert uniform shell designs and chromatography by Pedro Araujo; Steve Janagap (14-21).
► Review of the principles and application of Doehlert designs in chromatography. ► Critical discussion of the reported misconceptions regarding Doehlert designs. ► Estimation of the associated uncertainties generated by Doehlert designs.The principles of the Doehlert uniform shell designs (aka Doehlert designs) and their importance in the context of chromatography are discussed. The confidence of different models generated by Doehlert designs is studied by means of the uncertainty of the experimental points. The article provides an overview of analytical applications in chromatography with focus on single and coupled techniques and also discusses some reported blunders regarding Doehlert designs.
Keywords: Doehlert uniform shell designs; Uniform shell designs; Doehlert designs; Chromatography;

A review on second- and third-order multivariate calibration applied to chromatographic data by Juan A. Arancibia; Patricia C. Damiani; Graciela M. Escandar; Gabriela A. Ibañez; Alejandro C. Olivieri (22-30).
► Second- and third-order chromatographic analyses are reviewed. ► Multivariate calibration algorithms are described and discussed. ► Analytical chromametric applications are summarized.Quantitative analytical works developed by processing second- and third-order chromatographic data are reviewed. The various modes in which data of complex structure can be measured are discussed, with chromatographic separation providing either one or two of the data dimensions. This produces second-order data (matrices from uni-dimensional chromatography with multivariate detection or from two-dimensional chromatography) or third-order data (three-dimensional data arrays from two-dimensional chromatography with multivariate detection). The available algorithms for processing these data are classified and discussed, regarding their ability to cope with the ubiquitous phenomenon of retention time shifts from run to run. A summary of relevant works applying this combination of techniques is presented, with focus on quantitative analytical results. Special attention is paid to works achieving the full potentiality of the multidimensional data, i.e., the second-order advantage.
Keywords: Second- and third-order chromatographic data; Multivariate calibration; Second-order advantage;

Trends in data processing of comprehensive two-dimensional chromatography: State of the art by João T.V. Matos; Regina M.B.O. Duarte; Armando C. Duarte (31-45).
► Data processing of comprehensive 2D chromatography is a fast evolving subject. ► Concepts from 1D have been extended to 2D chromatography and this trend will persist. ► Algorithms are crucial for tweaking peak shift/shape and background/noise effects. ► MCR-ALS is the state-of-art technique for dealing with 4-way 2D chromatographic data.The operation of advanced chromatographic systems, namely comprehensive two-dimensional (2D) chromatography coupled to multidimensional detectors, allows achieving a great deal of data that need special care to be processed in order to characterize and quantify as much as possible the analytes under study. The aim of this review is to identify the main trends, research needs and gaps on the techniques for data processing of multidimensional data sets obtained from comprehensive 2D chromatography. The following topics have been identified as the most promising for new developments in the near future: data acquisition and handling, peak detection and quantification, measurement of overlapping of 2D peaks, and data analysis software for 2D chromatography. The rational supporting most of the data processing techniques is based on the generalization of one-dimensional (1D) chromatography although algorithms, such as the inverted watershed algorithm, use the 2D chromatographic data as such. However, for processing more complex N-way data there is a need for using more sophisticated techniques. Apart from using other concepts from 1D chromatography, which have not been tested for 2D chromatography, there is still room for new improvements and developments in algorithms and software for dealing with 2D comprehensive chromatographic data.
Keywords: Comprehensive two-dimensional chromatography; Multidimensional separations; Chemometrics; Peak detection; Resolution; Data treatment;

► Quantification of the flame-retardants PBDEs in tissues samples Salmo salar L. ► Telescoped extraction and purification of PBDEs using accelerated solvent extraction. ► Statistical design and optimization of the telescoped extraction and purification. ► Discovery of extractivity and recovery by means of multivariate design and analysis.The development of a multivariate study for a quantitative analysis of six different polybrominated diphenyl ethers (PBDEs) in tissue of Atlantic Salmo salar L. is reported. An extraction, isolation, and purification process based on an accelerated solvent extraction system was designed, investigated, and optimized by means of statistical experimental design and multivariate data analysis and regression. An accompanying gas chromatography–mass spectrometry analytical method was developed for the identification and quantification of the analytes, BDE 28, BDE 47, BDE 99, BDE 100, BDE 153, and BDE 154. These PBDEs have been used in commercial blends that were used as flame-retardants for a variety of materials, including electronic devices, synthetic polymers and textiles. The present study revealed that an extracting solvent mixture composed of hexane and CH2Cl2 (10:90) provided excellent recoveries of all of the six PBDEs studied herein. A somewhat lower polarity in the extracting solvent, hexane and CH2Cl2 (40:60) decreased the analyte %-recoveries, which still remain acceptable and satisfactory. The study demonstrates the necessity to perform an intimately investigation of the extraction and purification process in order to achieve quantitative isolation of the analytes from the specific matrix.
Keywords: Statistical experimental design; Response surface; Optimization; Principal component analysis; Polybrominated diphenyl ethers; Accelerated solvent extraction; Sample preparation; GC–MS;

► 15 mushroom species from 9 genera and 5 families of order Agaricales were analyzed. ► 31 fatty acids were measured as DMOX derivatives using GC–MS. ► Isomers 6-, 9- and 11–16:1 appeared to be characteristic for the examined species. ► PCA was used for multivariate exploration. ► Fatty acids separating the mushroom families were identified by PCA.Applying gas chromatography–mass spectrometry of 4,4-dimethyloxazoline fatty acid derivatives, the fatty acid composition of 15 mushroom species belonging to 9 genera and 5 families of order Agaricales growing in Bulgaria is determined. The structure of 31 fatty acids (not all present in each species) is unambiguously elucidated, with linoleic, oleic and palmitic acids being the main components (ranging between 70.9% (Marasmius oreades) and 91.2% (Endoptychum agaricoides)). A group of three hexadecenoic positionally isomeric fatty acids, 6-, 9- and 11–16:1, appeared to be characteristic components of the examined species. By applying chemometrics it was possible to show that the fatty acid composition closely reflects the classification of the species.
Keywords: Mushrooms; Agaricales; Fatty acids; DMOX derivatives; GC–MS; Chemometrics;

Similarity analyses of chromatographic fingerprints as tools for identification and quality control of green tea by G. Alaerts; J. Van Erps; S. Pieters; M. Dumarey; A.M. van Nederkassel; M. Goodarzi; J. Smeyers-Verbeke; Y. Vander Heyden (61-70).
► Evaluation of different similarity analysis parameters on green tea fingerprint profiles. ► Detection of non-genuine samples by similarity colour maps and evaluation plots. ► Adapted similarity score for the distinction between deviating and genuine samples. ► Dissimilar chromatographic tea fingerprints provided the same similarity assessment.Similarity assessment of complex chromatographic profiles of herbal medicinal products is important as a potential tool for their identification. Mathematical similarity parameters have the advantage to be more reliable than visual similarity evaluations of often subtle differences between the fingerprint profiles. In this paper, different similarity analysis (SA) parameters are applied on green-tea chromatographic fingerprint profiles in order to test their ability to identify (dis)similar tea samples. These parameters are either based on correlation or distance measurements. They are visualised in colour maps and evaluation plots. Correlation (r) and congruence (c) coefficients are shown to provide the same information about the similarity of samples. The standardised Euclidean distance (ds) reveals less information than the Euclidean distance (de), while Mahalanobis distances (dm) are unsuitable for the similarity assessment of chromatographic fingerprints. The adapted similarity score (ss*) combines the advantages of r (or c) and de. Similarity analysis based on correlation is useful if concentration differences between samples are not important, whereas SA based on distances also detects concentration differences well. The evaluation plots including statistical confidence limits for the plotted parameter are found suitable for the evaluation of new suspected samples during quality assurance. The ss* colour maps and evaluation plots are found to be the best tools (in comparison to the other studied parameters) for the distinction between deviating and genuine fingerprints. For all studied data sets it is confirmed that adequate data pre-treatment, such as aligning the chromatograms, prior to the similarity assessment, is essential. Furthermore, green-tea samples chromatographed on two dissimilar High-Performance Liquid Chromatography (HPLC) columns provided the same similarity assessment. Combining these complementary fingerprints did not improve the similarity analysis of the studied data set.
Keywords: Correlation and distance matrix; Fingerprint chromatography; Green tea; Quality control; Sample identification; Similarity analysis;

Quantification of blending of olive oils and edible vegetable oils by triacylglycerol fingerprint gas chromatography and chemometric tools by Cristina Ruiz-Samblás; Federico Marini; Luis Cuadros-Rodríguez; Antonio González-Casado (71-77).
► Reliable method for quantification of olive oils adulteration with vegetable oils. ► Determination triglycerides profile by HTGC–MS, use of all data points to get results. ► Chromatograms were pre-treated (baseline correction, peak alignment) before building models. ► Genetic algorithms were used as variable selection method to improve the models.A reliable procedure for the identification and quantification of the adulteration of olive oils in terms of blending with other vegetable oils (sunflower, corn, seeds, sesame and soya) has been developed. From the analytical viewpoint, the whole procedure relies only on the results of the determination of the triacylglycerol profile of the oils by high temperature gas chromatography–mass spectrometry. The chromatographic profiles were pre-treated (baseline correction, peak alignment using iCoshift algorithm and mean centering) before building the models. At first, a class-modeling approach, Soft Independent Modeling of Class Analogy (SIMCA) was used to identify the vegetable oil used blending. Successively, a separate calibration model for each kind of blending was built using Partial Least Square (PLS). The correlation coefficients of actual versus predicted concentrations resulting from multivariate calibration models were between 0.95 and 0.99. In addition, Genetic algorithms (GA–PLS), were used, as variable selection method, to improve the models which yielded R 2 values higher than 0.90 for calibration set. This model had a better predictive ability than the PLS without feature selection. The results obtained showed the potential of this method and allowed quantification of blends of olive oil in the vegetable oils tested containing at least 10% of olive oil.
Keywords: Olive oil; Vegetable oil; Blends; Genetic algorithm; PLS; GC–MS;

► Chemometric methods and chromatographic–spectroscopic data were applied. ► Partial least-squares regression was used to analyze overlapped enantiomer signals. ► Enantiomeric purity can be accurately determined without baseline enantioresolution. ► The R-(−)-ibuprofen was measured in the presence of 99.9% of the S-enantiomorph.In this report, we demonstrate how chiral liquid chromatography combined with multivariate chemometric techniques, specifically unfolded-partial least-squares regression (U-PLS), provides a powerful analytical methodology. Using U-PLS, strongly overlapped enantiomer profiles in a sample could be successfully processed and enantiomeric purity could be accurately determined without requiring baseline enantioresolution between peaks. The samples were partially enantioseparated with a permethyl-β-cyclodextrin chiral column under reversed-phase conditions. Signals detected with a diode-array detector within a wavelength range from 198 to 241 nm were recorded, and the data were processed by a second-order multivariate algorithm to decrease detection limits. The R-(−)-enantiomer of ibuprofen in tablet formulation samples could be determined at the level of 0.5 mg L−1 in the presence of 99.9% of the S-(+)-enantiomorph with relative prediction error within ±3%.
Keywords: High-performance liquid chromatography; Chiral analysis; Ibuprofen; Overlapped profiles; Multivariate calibration;

► Ant Colony Optimization, Relief, Stepwise regression and Genetic Algorithm as descriptor selection methods. ► SVM and MLR methods to model retention factors of the 83 diverse drugs. ► A slight preference may go to the ACO/SVMR model.A Quantitative Structure-Retention Relationship (QSRR) is proposed to estimate the chromatographic retention of 83 diverse drugs on a Unisphere poly butadiene (PBD) column, using isocratic elutions at pH 11.7. Previous work has generated QSRR models for them using Classification And Regression Trees (CART). In this work, Ant Colony Optimization is used as a feature selection method to find the best molecular descriptors from a large pool. In addition, several other selection methods have been applied, such as Genetic Algorithms, Stepwise Regression and the Relief method, not only to evaluate Ant Colony Optimization as a feature selection method but also to investigate its ability to find the important descriptors in QSRR. Multiple Linear Regression (MLR) and Support Vector Machines (SVMs) were applied as linear and nonlinear regression methods, respectively, giving excellent correlation between the experimental, i.e. extrapolated to a mobile phase consisting of pure water, and predicted logarithms of the retention factors of the drugs (log  k w). The overall best model was the SVM one built using descriptors selected by ACO.
Keywords: QSRR; Chromatographic retention; ACO; MLR; SVM; Relief method;

► We use a selection algorithm for rational test-set reduction. ► Test-set reduction will enhance the development of chiral screening strategies. ► A small but representative test-set will ease the update of these strategies. ► Reduction with 30% resulted in identical experimental conditions as initially defined.The last decades, many efforts have been made to design and develop chiral separation strategies for different analytical techniques. To ensure that these strategies are broadly applicable rather large test-sets of molecules with very diverse molecules are used. The most enantioselective and complementary separation systems are then used as screening conditions in separation strategies. Potential changes in conditions e.g. implementation of new chiral selectors, requires screening of the entire set to retain the most enantioselective systems. A rational reduction of the test-sets may open new perspectives for developing and updating separation strategies. In the present work, it is investigated whether the screening step of an existing separation strategy in polar organic solvents chromatography can be reconstructed based on reduced test-set results Therefore, the structures of the 58 molecules of the test-set are digitally drawn and their optimal geometrical conformations calculated. From these conformations 3D-molecular descriptors are calculated. The test-set reduction is performed using the Kennard and Stone algorithm: compounds with the most diverse descriptors are selected. The test-sets are gradually reduced with 10% starting from 90% to 30% of the initial size. The results pointed out that with some reduced test-sets the same chromatographic systems are selected. A test-set reduction with 30% (41 remaining compounds) seems possible without losing information on the global enantioselectivity and complementarity of the tested chiral stationary phases.
Keywords: Enantioseparations; Chiral separation strategies; Polar organic solvents chromatography; Test-set reduction;

Multivariate data analysis to evaluate the fingerprint peaks responsible for the cytotoxic activity of Mallotus species by C. Tistaert; G. Chataigné; B. Dejaegher; C. Rivière; N. Nguyen Hoai; M. Chau Van; J. Quetin-Leclercq; Y. Vander Heyden (103-113).
► Indication of peaks responsible for the cytotoxic activity of Mallotus species (O-PLS). ► Cytotoxic activity tests on cancerous and non-cancerous cell lines. ► Evaluation of the multivariate model prior and after alignment of the fingerprints. ► LC–MS analyses of the indicated peaks.The Mallotus genus comprises numerous species used as traditional medicines in oriental countries and provides scientists a broad basis in the search for pharmacologically active constituents. In this paper, the cytotoxicity of 39 Mallotus extracts, different in species, part of the plant used, origin, and harvest season, is evaluated combining cytotoxicity assays with fingerprint technology and data handling tools. At first, the antiproliferative activity of the plant extracts is analyzed both on a non-cancerous cell line (WI-38 – human lung fibroblast) and on a cancerous cell line (HeLa human cervix carcinoma). The results are linked to a data set of high-performance liquid chromatographic fingerprint profiles of the samples using multivariate calibration techniques. The regression coefficients of the multivariate model are then evaluated to indicate those peaks potentially responsible for the cytotoxic activity of the Mallotus extracts. In a final step, the cytotoxic extracts are analyzed by HPLC–MS and the indicated peaks identified.
Keywords: Fingerprints; Cytotoxicity; Multivariate calibration; Indication of peaks;

Potentially antioxidant compounds indicated from Mallotus and Phyllanthus species fingerprints by S. Thiangthum; B. Dejaegher; M. Goodarzi; C. Tistaert; A.Y. Gordien; N. Nguyen Hoai; M. Chau Van; J. Quetin-Leclercq; L. Suntornsuk; Y. Vander Heyden (114-121).
► Potentially antioxidant compounds indicated from fingerprints using PLS and O-PLS. ► Exploratory analysis discriminates active and non-active samples. ► Exploratory analysis discriminates between some species. ► Exploratory analysis indicates atypical samples.The genera of Mallotus and Phyllanthus contain several species that are commonly used as traditional medicines in oriental countries. Some species show interesting pharmaceutical activities, such as an antioxidant activity. To produce clinically useful medicines or food supplements (nutraceuticals) from these herbs, the species should be identified and a thorough quality control should be implemented. Nowadays, the integration of chromatographic and chemometric approaches allows a high-throughput identification and activity prediction of medicinal plants. In this study, Principal Component Analysis (PCA) and Hierarchical Cluster Analysis (HCA) were applied and compared to distinguish Mallotus and Phyllanthus species. Moreover, peaks from their chromatographic fingerprints, which were responsible for their antioxidant activity were assigned. For the latter purpose, the relevant information was extracted from the chromatographic fingerprints using linear multivariate calibration techniques, i.e., Partial Least Squares (PLS) and Orthogonal Projections to Latent Structures (O-PLS). Results reveal that exploratory analysis using PCA shows somewhat diverging clustering tendencies between Mallotus and Phyllanthus samples than HCA. However, both approaches mainly confirm each other. Concerning the multivariate calibration techniques, both PLS and O-PLS models demonstrate good predictive abilities. By comparing the regression coefficients of the models with the chromatographic fingerprints, the peaks that are potentially responsible for the antioxidant activity of the extracts could be confirmed.
Keywords: Chromatographic fingerprints; Antioxidant activity; Multivariate calibration; Mallotus species; Phyllanthus species;

► Using GC–MS and n-way techniques to quantify and identify triazines in one step. ► Fulfilling SANCO criteria of identification of triazines by GC–MS and PARAFAC. ► A procedure PARAFAC based for target identification with GC–MS. ► A new application of the second order property of PARAFAC in regulated analysis.The proposed procedure is described by applying it to develop an analytical method which fulfils the SANCO specifications. Nevertheless, the procedure would be valid for any other legal specification that requires the identification of the analyte by means its m/z values and retention time. To demonstrate the procedure, three herbicides (simazine, Sz; atrazine, Az; propazine, Pz), with terbuthylazine, Tz, as internal standard (IS) have been analysed by gas chromatography with mass spectrometry detection (GC/MS). The procedure consists of the following steps: (i) To record the data in the full scan mode (201 m/z ratios). (ii) To select four characteristic ions which make possible the unequivocal identification of each triazine according to the criteria established in the Document SANCO/12495/2011 by means of principal components and hierarchical clustering of variables; (iii) To build a calibration based on the PARAFAC decomposition with the data recorded in SIM mode at the four m/z ratios selected for each triazine. Afterwards several figures of merit have been evaluated. Bearing in mind that triazines are one of the most frequently used group of herbicides in agriculture and atrazine and simazine are included in the list of priority substances in Annex II of Directive 2008/105/EC, in this work, these analytes have been analysed in three natural waters. Prior to determination by gas chromatography with mass spectrometry detection (GC/MS) a step with solid phase extraction (SPE) has been carried out. The calibration set is made up of 40 standards 33 are external standards prepared in acetone and seven matrix matched prepared in deionised water subjected to the SPE procedure. Moreover, each kind of water, stream, well, and river, is analysed both unspiked and spiked. For the triazine determination, the second order PARAFAC advantage allows the use of samples prepared in acetone together with those prepared in deionised water subjected to SPE. The decision limit, CCα, and the capability of detection, CCβ, are calculated according to ISO 11843-2, assessing the false positive and false negative. The m/z ratios chosen fulfils the SANCO identification criteria and also the spectrum obtained in the PARAFAC decomposition, which is common in all samples for each triazine. However, when the same experimental data are used to carry out a univariate calibration with the abundance of the base peak of each triazines, a lot of samples lie outside the permitted tolerances depending on the reference experimental spectra used, despite the fact that all of them have a triazine content above the detection limit. Also, the PARAFAC calibration allows us to detect the test samples which are not similar to the calibration samples and in this way their mistaken quantification is avoided.
Keywords: GC–MS; Triazines; Parallel factor analysis; Detection capability; Decision limit; SANCO/12495/2011;

► 2-Bromophenol photodegradation is studied under UV and sunlight. ► UV spectroscopy, LC–DAD–MS and chemometrics are used for process interpretation. ► Multivariate Curve Resolution (MCR-ALS) has been the tool used for process modeling. ► MCR-ALS of LC–DAD–MS allowed modeling and identification of photoproducts. ► MCR-ALS applied to spectroscopic/chromatographic multiset elucidated the process mechanism.This work is focused on the study of the photodegradation of 2-bromophenol under the action of UV light and sunlight. The photodegradation process has been monitored using UV–Vis spectroscopy and High Performance Liquid Chromatography coupled to diode array and mass spectrometry detectors in tandem (HPLC–DAD–MS). Multivariate resolution methods, such as Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS) and hybrid soft- and hard-modeling-Multivariate Curve Resolution (HS-MCR), have been applied to the experimental data to obtain the information about the kinetic evolution and identification of the compounds involved in the photodegradation process. From the analysis of HPLC–DAD results, the complexity of the photodegradation process has been confirmed. Ten components were found to be involved in parallel, second- or higher-order reactions, which could not be ascertained from the spectroscopic results. The HPLC–MS results allowed postulating the identity of some of the compounds (such as hydroxyderivatives and bromophenol homologs) which resulted from the reactions of photohydrolysis, debromination and bromine transfer to different position of the phenol ring. The effect of the UV light and sunlight on the photodegradation process was found to affect mainly the rate of the reaction, but not the identity of the photoproducts formed. The advantages and limitations of the spectroscopic and chromatographic analysis were also discussed. The potential of combining spectroscopic and chromatographic data in a single multiset structure was also shown. This strategy, uses the advantage of the good definition of the process time axis from the spectroscopic experiment and the capability to distinguish among compounds, linked to the use of chromatographic information.
Keywords: Multivariate Curve Resolution-Alternating Least Squares (MCR-ALS); Hybrid soft- and hard-modeling-Multivariate Curve Resolution (HS-MCR); Photodegradation process; Process analysis;

A support vector machine-recursive feature elimination feature selection method based on artificial contrast variables and mutual information by Xiaohui Lin; Fufang Yang; Lina Zhou; Peiyuan Yin; Hongwei Kong; Wenbin Xing; Xin Lu; Lewen Jia; Quancai Wang; Guowang Xu (149-155).
► A new method was proposed to select the discriminative variables from the high dimension metabolome data. ► The developed method filters out noise and non-informative variables by means of artificial variables and mutual information. ► The discriminative variables were selected by SVM-RFE after removing noise. ► An better accuracy was obtained to distinguish among three liver diseases. ► 17 differential metabolites were identified to distinguish 3 liver diseases and the control.Filtering the discriminative metabolites from high dimension metabolome data is very important in metabolomics study. Support vector machine-recursive feature elimination (SVM-RFE) is an efficient feature selection technique and has shown promising applications in the analysis of the metabolome data. SVM-RFE measures the weights of the features according to the support vectors, noise and non-informative variables in the high dimension data may affect the hyper-plane of the SVM learning model. Hence we proposed a mutual information (MI)-SVM-RFE method which filters out noise and non-informative variables by means of artificial variables and MI, then conducts SVM-RFE to select the most discriminative features. A serum metabolomics data set from patients with chronic hepatitis B, cirrhosis and hepatocellular carcinoma analyzed by liquid chromatography–mass spectrometry (LC–MS) was used to demonstrate the validation of our method. An accuracy of 74.33 ± 2.98% to distinguish among three liver diseases was obtained, better than 72.00 ± 4.15% from the original SVM-RFE. Thirty-four ion features were defined to distinguish among the control and 3 liver diseases, 17 of them were identified.
Keywords: Artificial contrast variables; Mutual information; SVM-RFE; Liver diseases; Metabolomics;

Independent component analysis in non-hypothesis driven metabolomics: Improvement of pattern discovery and simplification of biological data interpretation demonstrated with plasma samples of exercising humans by Xiang Li; Jakob Hansen; Xinjie Zhao; Xin Lu; Cora Weigert; Hans-Ulrich Häring; Bente K. Pedersen; Peter Plomgaard; Rainer Lehmann; Guowang Xu (156-162).
► A novel approach based on descriptive statistics was established to optimize ICA model. ► ICA was elucidated as a novel method for the metabolic pattern analysis. ► Conclusive time dependent metabolic changes under exercise conditions were detected. ► ICA offers a novel perspective to elucidate key metabolite pattern.In a non-hypothesis driven metabolomics approach plasma samples collected at six different time points (before, during and after an exercise bout) were analyzed by gas chromatography–time of flight mass spectrometry (GC–TOF MS). Since independent component analysis (ICA) does not need a priori information on the investigated process and moreover can separate statistically independent source signals with non-Gaussian distribution, we aimed to elucidate the analytical power of ICA for the metabolic pattern analysis and the identification of key metabolites in this exercise study. A novel approach based on descriptive statistics was established to optimize ICA model. In the GC–TOF MS data set the number of principal components after whitening and the number of independent components of ICA were optimized and systematically selected by descriptive statistics. The elucidated dominating independent components were involved in fuel metabolism, representing one of the most affected metabolic changes occurring in exercising humans. Conclusive time dependent physiological changes of the metabolic pattern under exercise conditions were detected. We conclude that after optimization ICA can successfully elucidate key metabolite pattern as well as characteristic metabolites in metabolic processes thereby simplifying the explanation of complex biological processes. Moreover, ICA is capable to study time series in complex experiments with multi-levels and multi-factors.
Keywords: Independent component analysis; Metabolomics; Exercise; Metabolic profiling; GC–MS;

The design of microfluidic affinity chromatography systems for the separation of bioanalytes by Daniel Friedrich; Colin P. Please; Tracy Melvin (163-171).
► Simple approach for the design of microfluidic affinity chromatography devices. ► Multiple separation lanes and different surface-immobilised receptor patterns. ► Designs for the separation of large numbers of bioanalytes.The analytical (numerical) design of planar microfluidic affinity chromatography devices, which consist of multiple separation lanes and multiple, different surface-immobilised receptor patterns in each lane, is described. The model is based on the analytical solution of the transport-reaction equations in microfluidic systems of low Gratz number and for injection of small analyte plugs. The results reveal a simple approach for the design of microfluidic affinity chromatography devices tailored to the separation of bioanalytes, where receptors with high binding affinity are available. These devices have been designed for bioanalytical applications in mind, most notably for the proteomics field; the results are illustrated with an example using β-Amyloid binding peptides.
Keywords: Affinity chromatography; Microfluidic; mathematical modelling; separation; design strategy; β-Amyloid binding peptides;