What can go wrong at the data normalization step for identification of biomarkers?

Our study focuses on the removal of the so-called size effect, related to a different sample volume and/or concentration. This effect is associated with many types of instrumental signals, particularly with those originating from HPLC-DAD, LC-MS, and UPLC-MS. These signals do not carry any absolute information about the sample components. If the data comparison has to be performed based on sample fingerprints, then the size effect is undesired, and the shape effect is of main interest. With "shape", we refer to data information which is contained in the ratios between the variables. So far, different normalization methods have been applied to the removal of size effect. In our study, the performance of popular normalization methods is compared with those of the CODA (Compositional Data Analysis) methods, relying on log-ratio transformations, and the performance is evaluated through the prism of proper identification of biomarkers.

High-dimensional nested analysis of variance to assess the effect of production season, quality grade and steam pasteurization on the phenolic composition of fermented rooibos herbal tea.

A nested analysis of variance combined with simultaneous component analysis, ASCA, was proposed to model high-dimensional chromatographic data. The data were obtained from an experiment designed to investigate the effect of production season, quality grade and post-production processing (steam pasteurization) on the phenolic content of the infusion of the popular herbal tea, rooibos, at 'cup-of-tea' strength. Specifically, a four-way analysis of variance where the experimental design involves nesting in two of the three crossed factors was considered. For the purpose of the study, batches of fermented rooibos plant material were sampled from each of four quality grades during three production seasons (2009, 2010 and 2011) and a sub-sample of each batch was steam-pasteurized. The phenolic content of each rooibos infusion was characterized by high performance liquid chromatography (HPLC)-diode array detection (DAD). In contrast to previous studies, the complete HPLC-DAD signals were used in the chemometric analysis in order to take into account the entire phenolic profile. All factors had a significant effect on the phenolic content of a 'cup-of-tea' strength rooibos infusion. In particular, infusions prepared from the grade A (highest quality) samples contained a higher content of almost all phenolic compounds than the lower quality plant material. The variations of the content of isoorientin and orientin in the different quality grade infusions over production seasons are larger than the variations in the content of aspalathin and quercetin-3-O-robinobioside. Ferulic acid can be used as an indicator of the quality of rooibos tea as its content generally decreases with increasing tea quality. Steam pasteurization decreased the content of the majority of phenolic compounds in a 'cup-of-tea' strength rooibos infusion.

FTIR spectroscopic characterization of differently cultivated food related yeasts.

The application of Fourier Transform Infrared Spectroscopy for characterization of yeasts is growing rapidly. Since it is known that the phenotypic expression of yeast cells depends sensitively on the nutrients that are available in the growth medium, one standardized growth medium is usually used for identification and characterization purposes in order to obtain reproducible FTIR signals. Since our recently developed high-throughput micro-cultivation protocol has the capacity to use more than one standardized growth medium, we wanted to investigate if the parallel use of multiple growth media can improve identification results. For this purpose, five different cultivation media (YP, YPD, YMB, SAB and SD) were used. In total 91 food spoilage yeast strains of 12 different genera were cultivated in different cultivation media and subsequently characterized by FTIR spectroscopy. For spectral identifications, Radial Basis Function-Partial Least Squares (RBF-PLS) was used in combination with cross-model validation where an inner cross-validation loop was used to optimize the model, while in an outer loop an independent test set was kept aside to test the optimized model. Sensitivity and specificity were evaluated for each studied genus class. The results show that the YMB selective medium gave the best discrimination results for 9 of the 12 genera with sensitivity above 90%. Only three genera showed better identification results on other media (Clavispora and Metschnikowia on medium SD, Debaryomyces on medium YPD). We therefore suggest to use the media SD, YPD in combination with the YMB medium for the identification of food spoilage yeasts.

Simultaneous determination of Solvent Yellow 124 and Solvent Red 19 in diesel oil using fluorescence spectroscopy and chemometrics.

Differences in tax levels for diesel oil stimulate the illegal removal of characteristic diazo compounds purposely added to designate its possible usage. In order to reduce the losses in the national income, there is a strong need to develop a sensitive and cost-effective analytical procedure for the detection of this illegal action. In this study, we describe a novel analytical approach for a qualitative and quantitative determination of two diazo compounds (Solvent Yellow 124 and Solvent Red 19) that are usually added to diesel oil. The methodology proposed combines the use of excitation-emission matrix fluorescence spectroscopy as an analytical technique and partial least squares regression as a multiple modeling tool. With this new methodology, relatively low root mean square errors of prediction (for independent set of test samples) that are equal to 0.223 for Solvent Red 19 and 0.263 for Solvent Yellow 124, were obtained and the results were stable, which were indicated by an analysis performed after 48 and 96 h. The methodology is also nondestructive and allows for (i) simultaneous detection of diesel oil additives, (ii) determination of satisfactory limits of detection (0.048 and 0.042 mg L(-1) for Solvent Red 19 and Solvent Yellow 124, respectively), and (iii) obtaining of considerably low relative standard deviations of 2.33% for Solvent Yellow 124 and of 3.23% for Solvent Red 19 in comparison with the existing norm level.

Interpretation of analysis of variance models using principal component analysis to assess the effect of a maternal anticancer treatment on the mineralization of rat bones.

The goal of the present study is to assess the effects of anticancer treatment with cyclophosphamide and cytarabine during pregnancy on the mineralization of mandible bones in 7-, 14- and 28-day-old rats. Each bone sample was described by its X-ray fluorescence spectrum characterizing the mineral composition. The data collected are multivariate in nature and their structure is difficult to visualize and interpret directly. Therefore, methods like analysis of variance-principal component analysis (ANOVA-PCA) and ANOVA-simultaneous component analysis (ASCA), which are suitable for the analysis of highly correlated spectral data and are able to incorporate information about the underlined experimental design, are greatly valued. In this study, the ASCA methodology adapted for unbalanced data was used to investigate the impact of the anticancer drug treatment during pregnancy on the mineralization of the mandible bones of newborn rats and to examine any changes in the mineralization of the bones over time. The results showed that treatment with cyclophosphamide and cytarabine during pregnancy induces a decrease in the K and Zn levels in the mandible bones of newborns. This suppresses the development of mandible bones in rats in the early stages (up to 14 days) of formation. An interesting observation was that the levels of essential minerals like K, Mg, Na and Ca vary considerably in the different regions of the mandible bones.

Methods for the exploratory analysis of two-dimensional chromatographic signals.

In this article several approaches for the exploratory analysis of two-dimensional chromatographic signals (fingerprints) are presented. Their usefulness is illustrated on experimental chromatographic data obtained from high performance liquid chromatography using the photodiode-array detector (HPLC-DAD). Among the methods discussed are principal component analysis (PCA), hierarchical clustering methods and several N-way techniques such as PARAFAC, PARAFAC2 and Tucker3. In addition to the N-way methods, other approaches that allow for comparing samples represented by two-dimensional fingerprints are also presented (the Rv coefficient, the STATIS approach and 'fuzzy' variants of the similarity matrix). Exploratory analysis of the HPLC-DAD data with peak shifts is also discussed.

Relating gas chromatographic profiles to sensory measurements describing the end products of the Maillard reaction.

Often in analytical practice, a set of samples is described by different types of measurements in the hope that a comprehensive characterisation of samples will provide a more complete picture and will help in determining the similarities among samples. The main focus is then on how to combine the information described by different measurement variables and how to analyse it simultaneously. In other words, the main goal is to find a common representation of samples that emphasises the individual and common properties of the different blocks of variables. Several methods can be adopted for the simultaneous analysis of multiblock data with a common object mode. These are: consensus principal component analysis (CPCA), SUM-PCA, multiple factor analysis (MFA) and structuration des tableaux à trois indices de la statistique (STATIS).In this article we present a comparison of the performances of these methods for data describing the chemistry and sensory profiles of the Maillard reaction products. The aroma compounds formed during the reaction of thermal heating between one or two selected amino acids and one or two reducing sugars have been analysed by head space gas chromatography and the intensity and nature of the odour of the resulting products has been evaluated according to selected descriptors by a panel of sensory experts.The results showed that using the information of the chromatographic and sensory data in conjunction enhanced the interpretability of the data. SUM-PCA and more specifically multiple factor analysis, MFA, allowed for a detailed study of the similarities of mixtures in terms of reaction products and sensory profiles.

Automated alignment of one-dimensional chromatographic fingerprints.

A general framework for the automatic alignment of one-dimensional chromatographic signals is presented in this article. The alignment of signals was achieved by explicitly modeling the warping function. Its shape was estimated using a linear combination of several B-spline functions. The coefficients of the spline functions were found in the course of an optimization procedure to maximize the Pearson's correlation coefficient between a target chromatogram and aligned chromatogram(s). The computational requirements of the method are discussed with respect to the correlation optimized warping method, frequently used for the alignment of chromatographic signals. As illustrated with two sets of one-dimensional chromatographic fingerprints, the automatic alignment approach performs well even when non-linear peak shifts need to be corrected. It can be applied in an on-the-fly manner since the alignment of signals is rapid.

Near-infrared reflectance spectroscopy and multivariate calibration techniques applied to modelling the crude protein, fibre and fat content in rapeseed meal.

Near-infrared reflectance spectroscopy (NIRS) is often applied when a rapid quantification of major components in feed is required. This technique is preferred over the other analytical techniques due to the relatively few requirements concerning sample preparations, high efficiency and low costs of the analysis. In this study, NIRS was used to control the content of crude protein, fat and fibre in extracted rapeseed meal which was produced in the local industrial crushing plant. For modelling the NIR data, the partial least squares approach (PLS) was used. The satisfactory prediction errors were equal to 1.12, 0.13 and 0.45 (expressed in percentages referring to dry mass) for crude protein, fat and fibre content, respectively. To point out the key spectral regions which are important for modelling, uninformative variable elimination PLS, PLS with jackknife-based variable elimination, PLS with bootstrap-based variable elimination and the orthogonal partial least squares approach were compared for the data studied. They enabled an easier interpretation of the calibration models in terms of absorption bands and led to similar predictions for test samples compared to the initial models.

Raman spectroscopy as a process analytical technology (PAT) tool for the in-line monitoring and understanding of a powder blending process.

The aim of this study is to propose a strategy to implement a PAT system in the blending step of pharmaceutical production processes. It was examined whether Raman spectroscopy can be used as PAT tool for the in-line and real-time endpoint monitoring and understanding of a powder blending process. A screening design was used to identify and understand the significant effects of two process variables (blending speed and loading of the blender) and of a formulation variable (concentration of active pharmaceutical ingredient (API): diltiazem hydrochloride) upon the required blending time (response variable). Interactions between the variables were investigated as well. A Soft Independent Modelling of Class Analogy (SIMCA) model was developed to determine the homogeneity of the blends in-line and real-time using Raman spectroscopy in combination with a fiber optical immersion probe. One blending experiment was monitored using Raman and NIR spectroscopy simultaneously. This was done to verify whether two independent monitoring tools can confirm each other's endpoint conclusions. The analysis of the experimental design results showed that the measured endpoints were excessively rounded due to the large measurement intervals relative to the first blending times. This resulted in effects and critical effects which cannot be interpreted properly. To be able to study the effects properly, the ratio between the blending times and the measurement intervals should be sufficiently high. In this study, it anyway was demonstrated that Raman spectroscopy is a suitable PAT tool for the endpoint control of a powder blending process. Raman spectroscopy not only allowed in-line and real-time monitoring of the blend homogeneity, but also helped to understand the process better in combination with experimental design. Furthermore, the correctness of the Raman endpoint conclusions was demonstrated for one process by using a second independent endpoint monitoring tool (NIR spectroscopy). Hence, the use of two independent techniques for the control of one response variable not only means a mutual confirmation of both methods, but also provides a higher certainty in the determined endpoint.

Classification of data with missing elements and outliers.

Missing elements and outliers can often occur in experimental data. The presence of outliers makes the evaluation of any least squares model parameters difficult, while the missing values influence the adequate identification of outliers. Therefore, approaches that can handle incomplete data containing outliers are highly valued. In this paper, we present the expectation-maximization robust soft independent modeling of class analogy approach (EM-S-SIMCA) based on the recently introduced spherical SIMCA method. Several important issues like the possibility of choosing the complexity of the model with the leverage correction procedure, the selection of training and test sets using methods of uniform design for incomplete data and prediction of new samples containing missing elements are discussed. The results of a comparison study showed that EM-S-SIMCA outperforms the classic expectation-maximization SIMCA method. The performance of the method was illustrated on simulated and real data sets and led to satisfactory results.

No-alignment-strategies for exploring a set of two-way data tables obtained from capillary electrophoresis-mass spectrometry.

Hyphenated techniques such as capillary electrophoresis-mass spectrometry (CE-MS) or high-performance liquid chromatography with diode array detection (HPLC-DAD), etc., are known to produce a huge amount of data since each sample is characterized by a two-way data table. In this paper different ways of obtaining sample-related information from a set of such tables are discussed. Working with original data requires alignment techniques due to time shifts caused by unavoidable variations in separation conditions. Other pre-processing techniques have been suggested to facilitate comparison among samples without prior peak alignment, for example, 'binning' and/or 'blurring' the data along the time dimension. All these techniques, however, require optimization of some parameters, and in this paper an alternative parameter-free method is proposed. The individual data tables (X) are represented as Gram matrices (XXT), where the summation is taken over the time dimension. Hence the possible variations in time scale are eliminated, while the time information is at least partly preserved by the correlation structure between the detection channels. For comparison among samples, a similarity matrix is constructed and explored by principal component analysis and hierarchical clustering. The Gram matrix approach was tested and compared to some other methods using 'binned' and 'blurred' data for a data set with CE-MS runs on urine samples. In addition to data exploration by principal component analysis and hierarchical clustering, a discriminant partial least squares model was constructed to discriminate between the samples that were taken with and without the prior intake of a drug. The result showed that the proposed method is at least as good as the others with respect to cluster identification and class prediction. A distinct advantage is that there is no need for parameter optimization, while a potential drawback is the large size of the Gram matrices for data with high mass resolution.

Target selection for alignment of chromatographic signals obtained using monochannel detectors.

The alignment of instrumental signals, such as chromatograms, is regarded as an important step before applying multivariate chemometric techniques for data analysis. Nowadays, many alignment techniques are available and they differ in achieving their goal. They can correct peak shifts in a set of chromatograms with differing degrees of success. Almost all alignment techniques, with few exceptions [e.g., W. Yu, B. Wu, N. Lin, K. Stone, K. Williams, H. Zhao, Comput. Biol. Chem. 30 (2006) 27], require a careful choice of the target profile. The selection of a target signal is not an easy task and some difficulties related to this selection are discussed in this paper. An analysis of several simulated sets of chromatographic signals showed that the target selection can be a crucial step if the aligned signals are then used as input to unsupervised approaches, such as, e.g., principal component analysis and to supervised methods like discriminant partial least squares. Different proposals for target selection known-to-date are reviewed. As demonstrated in our study the target profile with the highest correlation coefficient among all the signals studied gave the most satisfactory results.

Robust partial least squares model for prediction of green tea antioxidant capacity from chromatograms.

In this paper a robust version of the partial least squares model (partial robust M-regression, PRM) was built to predict the total antioxidant capacity of green tea extracts. In order to construct a calibration model, chromatograms obtained by a fast high-performance liquid chromatographic method on a monolithic silica column were related with the total antioxidant capacity of green tea extracts as determined by the Trolox antioxidant capacity method. Since natural samples are the subject of the study, some outlying samples are present in the data, as shown in an earlier work. Therefore, to construct reliable calibration models, they were detected and removed prior to modeling. With the applied robust partial least squares approach, where a weighting scheme is embedded to down-weight the negative influence of outliers upon the model it is possible to construct a robust calibration model, without prior identification of outlying objects. It was shown that a robust model, allowing satisfactory prediction for test samples, can be used in controlling green tea antioxidant capacity based on their chromatograms. The constructed robust partial least squares model was shown to have virtually the same fit and predictive power as the classical partial least squares model when outlying samples were removed from the data.

Start-to-end processing of two-dimensional gel electrophoretic images.

Gel electrophoresis serves as a basic analytical tool in the proteomic studies. However, processing of gel electrophoretic images is still the main bottleneck of data analysis, and there is an increasing need for the fully automated approaches. The proposed start-to-end strategy of analyzing the gel images consists of chemometric tools, which allow their effective preprocessing, automatic warping, and data modeling. The image preprocessing techniques: denoising in the wavelet domain and the penalized asymmetric least squares approach for the background estimation are proposed. Matching of images is based on fuzzy warping of features, extracted from the gel images. For the classification or calibration purpose, multivariate approaches such, as partial least squares (PLS) or kernel-PLS methods are used. Performance of the proposed strategy is demonstrated on the real set of the two-dimensional gel images.

Dealing with missing values and outliers in principal component analysis.

An efficient methodology for dealing with missing values and outlying observations simultaneously in principal component analysis (PCA) is proposed. The concept described in the paper consists of using a robust technique to obtain robust principal components combined with the expectation maximization approach to process data with missing elements. It is shown that the proposed strategy works well for highly contaminated data containing different amounts of missing elements. The authors come to this conclusion on the basis of the results obtained from a simulation study and from analysis of a real environmental data set.

Chemometric analysis of the water purification process data.

The aim of this work was to show usefulness of chemometric analysis in processing of the data describing production of drinking water in the Silesian region of Poland. Water samples have been collected within the period of 1 year and the quality of water was characterized by a number of physical, chemical and microbiological parameters. Principal component analysis (PCA) and STATIS (Structuration des Tableaux A Trois Indices de la Statistique) were employed to obtain the knowledge about the complete water treatment process. PCA makes it possible to uncover seasonal changes influencing the water treatment process. In particular, it was found out that the salt content, hardness and conductivity of water tend to obtain higher levels in winter rather than in summer, and the relatively lower acidity is also to be expected in winter. The sensory quality of water is considerably improved over the consecutive purification steps. Complementary information about the individual technological units of the process is gained with the STATIS approach. The obtained results show that the water produced by the two independent filtering branches of the water plant is of similar quality and the prescribed quality characteristics of drinking water are fulfilled.

Improving QSAR models for the biological activity of HIV Reverse Transcriptase inhibitors: Aspects of outlier detection and uninformative variable elimination.

The goal of this study is to derive a methodology for modeling the biological activity of non-nucleoside HIV Reverse Transcriptase (RT) inhibitors. The difficulties that were encountered during the modeling attempts are discussed, together with their origin and solutions. With the selected multivariate techniques: robust principal component analysis, partial least squares, robust partial least squares and uninformative variable elimination partial least squares, it is possible to explore and to model the contaminated data satisfactory. It is shown that these techniques are versatile and valuable tools in modeling and exploring biochemical data.

Classification and regression trees--studies of HIV reverse transcriptase inhibitors.

In this paper, the application of Classification And Regression Trees (CART) is presented for the analysis of biological activity of Non-Nucleoside Reverse Transcriptase Inhibitors (NNRTIs). The data consist of the biological activities, expressed as pIC50, of 208 NNRTIs against wild-type HIV virus (HIV-1) and four mutant strains (181C, 103N, 100I, 188L) and the computed interaction energies with the Reverse Transcriptase (RT) binding pocket. CART explains the observed biological activity of NNRTIs in terms of interactions with individual amino acids in the RT binding pocket, i.e., the original data variables.