Imaging live cells grown on a three dimensional collagen matrix using Raman microspectroscopy

Three dimensional collagen gels have been used as matrices for the imaging of live cells by Raman spectroscopy. The study is conducted on a human lung adenocarcinoma (A549) and a spontaneously immortalized human epithelial keratinocyte (HaCaT) cell line. The lateral resolution of the system has been estimated to be <1.5 μm making it possible to access the subcellular organization. Using K-means clustering analysis, it is shown that the different subcellular compartments of individual cells can be identified and differentiated. The biochemical specificity of the information contained in the Raman spectra allows the visualization of differences in the molecular signature of the different sub-cellular structures. Furthermore, to enhance the chemical information obtained from the spectra, principal component analysis has been employed, allowing the identification of spectral windows with a high variability. The comparison between the loadings calculated and spectra from pure biochemical compounds enables the correlation of the variations observed with the molecular content of the different cellular compartments.     

Blind image analysis for the compositional and structural characterization of plant cell walls

A new image analysis strategy is introduced to determine the composition and the structural characteristics of plant cell walls by combining Raman microspectroscopy and unsupervised data mining methods. The proposed method consists of three main steps: spectral preprocessing, spatial clustering of the image and finally estimation of spectral profiles of pure components and their weights. Point spectra of Raman maps of cell walls were preprocessed to remove noise and fluorescence contributions and compressed with PCA. Processed spectra were then subjected to k-means clustering to identify spatial segregations in the images. Cell wall images were reconstructed with cluster identities and each cluster was represented by the average spectrum of all the pixels in the cluster. Pure components spectra were estimated by spectral entropy minimization criteria with simulated annealing optimization. Two pure spectral estimates that represent lignin and carbohydrates were recovered and their spatial distributions were calculated. Our approach partitioned the cell walls into many sublayers, based on their composition, thus enabling composition analysis at subcellular levels. It also overcame the well known problem that native lignin spectra in lignocellulosics have high spectral overlap with contributions from cellulose and hemicelluloses, thus opening up new avenues for microanalyses of monolignol composition of native lignin and carbohydrates without chemical or mechanical extraction of the cell wall materials.     

Cluster analysis of infrared spectra of rabbit cortical bone samples during maturation and growth

Bone consists of an organic and an inorganic matrix. During development, bone undergoes changes in its composition and structure. In this study we apply three different cluster analysis algorithms [K-means (KM), fuzzy C-means (FCM) and hierarchical clustering (HCA)], and discriminant analysis (DA) on infrared spectroscopic data from developing cortical bone with the aim of comparing their ability to correctly classify the samples into different age groups. Cortical bone samples from the mid-diaphysis of the humerus of New Zealand white rabbits from three different maturation stages (newborn (NB), immature (11 days-1 month old), mature (3-6 months old)) were used. Three clusters were obtained by KM, FCM and HCA methods on different spectral regions (amide I, phosphate and carbonate). The newborn samples were well separated (71-100% correct classifications) from the other age groups by all bone components. The mature samples (3-6 months old) were well separated (100%) from those of other age groups by the carbonate spectral region, while by the phosphate and amide I regions some samples were assigned to another group (43-71% correct classifications). The greatest variance in the results for all algorithms was observed in the amide I region. In general, FCM clustering performed better than the other methods, and the overall error was lower. The discriminate analysis results showed that by combining the clustering results from all three spectral regions, the ability to predict the correct age group for all samples increased (from 29-86% to 77-91%). This study is the first to compare several clustering methods on infrared spectra of bone. Fuzzy C-means clustering performed best, and its ability to study the degree of memberships of samples to each cluster might be beneficial in future studies of medical diagnostics.     

Hyphenation of column liquid chromatography and Raman spectroscopy via a liquid-core waveguide: chemometrical elimination of spectral eluent background

In column liquid chromatography (LC) coupled to conventional Raman spectroscopy (RS) removal of the spectral background of the eluent is often demanding, because of the strong signals of the organic modifier. A new chemometrical method is proposed, called the eluent background subtraction (EBS) method, which can correct for small shape and intensity differences of the eluent spectra. The variations in the eluent spectra are modelled using principal component analysis (PCA). The PCA loading vectors are subsequently used for eluent background correction of the elution spectra of the analyte. The loading vectors are fitted under these spectra by an asymmetric least-squares method. This method was successfully applied under various experimental conditions and performed much better than conventional background correction methods. Analyte detectability was improved by (weighted) averaging of all elution spectra and smoothing via a p-spline function.     

Fourier-transform infrared (FTIR) spectroscopy for monitoring and determining the degree of crystallisation of polyhydroxyalkanoates (PHAs)

FTIR spectroscopy has been used to monitor and determine the degree of crystallisation in a sample of polyhydroxybutyrate-co-14%valerate (PHB–co-14%HV). Time series spectra of solution-cast films of the polymer revealed spectral changes attributed to the onset of crystallisation. Curve fitting was used to obtain an absolute measure of crystallinity. Mean centred principal-component analysis (PCA) revealed that 99.9% of the spectral variance could be attributed to factor 1. The loadings plot for factor 1 contained features attributable to crystalline and amorphous phases. These features were opposite in sign, indicating that changes in the spectra with the onset of crystallisation are simultaneous and opposite in direction, i.e. as the crystalline band increases the amorphous band decreases. Cross-peaks in asynchronous 2D correlation maps indicate there are likely to be very minor components that are changing out of phase. The presence of these minor components is supported by examination of the loadings of higher factors in the PCA model. PCA has been shown to be suitable for determining the number of dynamic spectral features and has enabled relative and objective monitoring of crystallisation kinetics.     

Eigenvalues and principal component loadings of heavily overlapped vibrational spectra

In order to illustrate the possibilities of principal component analysis in determining the number of components in a system of heavily overlapped spectra, several numerical spectral models were formed of bands with very close parameters. The models consisted of three bands, whose peak positions were locally shifted and noise added. For all the cases the relations between eigenvalues and principal component loadings were considered. It was shown that for those complex spectra for which peak positions and band halfwidths can be determined with high accuracy, eigenvalues criteria could easily indicate the number of components. For all analyzed models, the consideration of the shape of loadings was proven to have a high importance. In limited number of cases the shape of a loading can make the results of eigenvalue analysis more understandable. It has been shown that the noise can be treated as a main limitation in the application of the method to this type of the spectra.     

Studies of chemical fixation effects in human cell lines using Raman microspectroscopy

The in vitro study of cellular species using Raman spectroscopy has proven a powerful non-invasive modality for the analysis of cell constituents and processes. This work uses micro-Raman spectroscopy to study the chemical fixation mechanism in three human cell lines (normal skin, normal bronchial epithelium, and lung adenocarcinoma) employing fixatives that preferentially preserve proteins (formalin), and nucleic acids (Carnoy’s fixative and methanol–acetic acid). Spectral differences between the mean live cell spectra and fixed cell spectra together with principal components analysis (PCA), and clustering techniques were used to analyse and interpret the spectral changes. The results indicate that fixation in formalin produces spectral content that is closest to that in the live cell and by extension, best preserves the cellular integrity. Nucleic acid degradation, protein denaturation, and lipid leaching were observed with all fixatives and for all cell lines, but to varying degrees. The results presented here suggest that the mechanism of fixation for short fixation times is complex and dependent on both the cell line and fixative employed. Moreover, important spectral changes occur with all fixatives that have consequences for the interpretation of biochemical processes within fixed cells. The study further demonstrates the potential of vibrational spectroscopy in the characterization of complex biochemical processes in cells at a molecular level.     

Efficient and sample-specific interpretation of ToF-SIMS data by additional postprocessing of principal component analysis results

Time-of-flight secondary ion mass spectrometry (ToF-SIMS) is a powerful tool for surface analysis, but fragmentation of molecular species during the SIMS process may lead to complex mass spectra. While the fragmentation pattern is typically characteristic for each compound, industrial samples are engineered materials, and, thus, may contain a mixture of many compounds, which may result in a variety of overlapping peak patterns in ToF-SIMS spectra. Consequently, the process of data evaluation is challenging and time-consuming. Principal component analysis (PCA) can be used to simplify data analysis for complex sample systems. Especially, correlation loadings were observed as an ideal tool to identify relevant signals in PCA results, which induce the separation of different sample groups. This is because correlation loadings show the relevance of signals independent from their intensity in the raw data. In correlation loadings, however, fragmentation patterns are no longer observed and the identification of peaks' sum formulas is challenging. In this study, a new approach is presented, which simplifies peak identification and assignment in ToF-SIMS spectra after PCA is performed. The approach uses a mathematical transformation that projects PCA results, in particular loadings and correlation loadings, in the direction of specific sample groups. The approach does not change PCA results but rather presents them in a new way. This method allows to visualize characteristic spectra for specific sample groups that contain only relevant signals and, additionally, visualize fragmentation patterns. Data analysis is simplified and helps the user to focus on data interpretation rather than processing.     

A robust clustering approach for NMR spectra of natural product extracts

A robust method was developed to cluster similar NMR spectra from partially purified extracts obtained from a range of marine sponges and a plant biota. The NMR data were acquired using microtiter plate NMR (VAST) in protonated solvents. A sample data set which contained several clusters was used to optimize the protocol. The evaluation of the robustness was performed using three different clustering methods: tree clustering analysis, K-means clustering and multidimensional scaling. These methods were compared for consistency using the sample data set and the optimized methodology was applied to clustering of a set of spectra from partially purified biota extracts.     

K-means Find Density Peaks in Molecular Conformation Clustering

Performing cluster analysis on molecular conformation is an important way to find the representative conformation in the molecular dynamics trajectories. Usually, it is a critical step for interpreting complex conformational changes or interaction mechanisms. As one of the density-based clustering algorithms, find density peaks (FDP) is an accurate and reasonable candidate for the molecular conformation clustering. However, facing the rapidly increasing simulation length due to the increase in computing power, the low computing efficiency of FDP limits its application potential. Here we propose a marginal extension to FDP named K-means find density peaks (KFDP) to solve the mass source consuming problem. In KFDP, the points are initially clustered by a high efficiency clustering algorithm, such as K-means. Cluster centers are defined as typical points with a weight which represents the cluster size. Then, the weighted typical points are clustered again by FDP, and then are refined as core, boundary, and redefined halo points. In this way, KFDP has comparable accuracy as FDP but its computational complexity is reduced from O\begin{document}$(n^2)$\end{document} to O\begin{document}$(n)$\end{document}. We apply and test our KFDP method to the trajectory data of multiple small proteins in terms of torsion angle, secondary structure or contact map. The comparing results with K-means and density-based spatial clustering of applications with noise show the validation of the proposed KFDP.     

Classification of Edible Oils by Infrared Spectroscopy with Optimized K-Means Clustering by a Hybrid Particle Swarm Algorithm

The K-means algorithm has some limitations including dead-unit properties, heavy dependence on the initial choice of cluster centers, convergence to local optima, and sensitivity to the number of clusters. This paper presents an efficient algorithm that optimizes K-means clustering by a hybrid particle swarm algorithm. The modified discrete algorithm is used to select variables and is continuously applied to update cluster centers simultaneously. The nearest center classification is then employed to classify the test samples. The proposed algorithm was applied to discriminate various edible oil varieties by employing Fourier transform infrared spectroscopy. As a comparison, the common K-means clustering, principal component analysis, and partial least squares techniques were also applied to classify these edible oil samples. Results demonstrated that the proposed method is an accurate and rapid strategy for identifying edible oils.     

Raman spectral imaging of single cancer cells: probing the impact of sample fixation methods

Raman spectroscopy has proven its potential for the analysis of cell constituents and processes. However, sample preparation methods compatible with clinical practice must be implemented for collection of accurate spectral information. This study aims at assessing, using micro-Raman imaging, the effects of some routinely used fixation methods such as formalin-fixation, formalin-fixation/air drying, cytocentrifugation, and air drying on intracellular spectral information. Data were compared with those acquired from single living cells. In parallel to these spectral information, cell morphological modifications that accompany sample preparation were compared. Spectral images of isolated cells were first analyzed in an unsupervised way using hierarchical cluster analysis (HCA), which allowed delimitation of the cellular compartments. The resulting nuclei cluster centers were compared and revealed at the molecular level that fixation induced changes in spectral information assigned to nucleic acids and proteins. In a second approach, a supervised fitting procedure using model spectra of DNA, RNA, and proteins, chemically extracted from living cells, revealed very small modifications at the level of the localization and quantification of these macromolecules. Finally, HCA and principal components analysis (PCA) performed on individual spectra randomly selected from the nuclear regions showed that formalin-fixation and cytocentrifugation are sample preparation methods that have little impact on the biochemical information as compared to living conditions. Any step involving cell air drying seems to accentuate the spectral deviations from the other preparation methods. It is therefore important in a future context of spectral cytology to take into account these variations.     

Screening phosphatidylcholine biomarkers in mouse liver extracts from a hypercholesterolemia study using ESI-MS and chemometrics

When fed a high-fat, high-cholesterol diet (HFD), homozygous LDL receptor knockout mice exhibit extremely high levels of plasma cholesterol that are expected to influence liver metabolism. One step in the investigation of potential hepatic alterations was the analysis of organic extracts of livers from these and control mice by electrospray mass spectrometry (ESI-MS). Chemometrics (bioinformatics) analysis shows that the sample spectra cluster into two groups: one from mice with plasma cholesterol levels in excess of 900 mg dL⁻¹ and one from animals with cholesterol levels of 60–250 mg dL⁻¹. The loadings plot of the first PC in the principal-components analysis (PCA) reveals the chemical basis for clustering, i.e., biomarkers present at different concentrations in the different groups. The exact masses of the key peaks in this loadings plot indicate these species are phosphatidylcholines (PtdChos). This assignment is confirmed by tandem MS. Partial least-squares (PLS) with variable selection shows that the spectra are well correlated with plasma total cholesterol, HDL cholesterol, and triglyceride (TG) levels.     

Use of multivariate analysis of MIR spectra to study bread staling

Different kinds of bread, stored at constant temperature and at controlled humidity conditions for a week since their manufacturing date, were analysed by Attenuated Total Reflectance-Fourier Transform InfraRed (ATR-FTIR) spectroscopy. The collected spectra were processed by Principal Component Analysis (PCA), in order to evaluate the changes occurring during bread ageing. For the sake of comparison, the 1060-950 cm spectral window has been also investigated by curve-fitting methods. It was observed that the first PC increases monotonically with ageing of samples. Furthermore, the more influential variables on PC1 correspond to spectral regions where are located stretching and bending bands, which are mainly attributed to typical starch bonds vibrations.     

Combination of FTIR spectral imaging and chemometrics for tumour detection from paraffin-embedded biopsies

FTIR spectral imaging was applied on formalin-fixed paraffin-embedded biopsies from colon and skin cancerous lesions. These samples were deposited onto different substrates (zinc selenide and calcium fluoride respectively) and embedded using two types of paraffin. Formalin fixation followed by paraffin embedding is the gold standard in tissue storage. It can preserve molecular structures and it is compatible with immunohistochemistry. However, paraffin absorption bands are significant in the mid-infrared region and can mask some molecular vibrations of the tissue. Direct data processing was applied on spectral images without any chemical dewaxing of the tissues. Extended Multiplicative Signal Correction was used to correct the spectral contribution from paraffin. For this purpose, the signal of paraffin was modelled using Principal Component Analysis and paraffin spectra were removed from the raw images based on an outlier detection. Then, pseudo-colour images were computed by K-means clustering in order to highlight histological structures of interest. This robust chemometrics methodology was applied on the two samples. Tumour areas were successfully demarcated from the rest of the tissue in both colon and skin independently of the embedding material and of the substrate.     

Spectrally Resolved Imaging of Cabot Rings and Howell-Jolly Bodies

The spectral characteristics of erythropoietic cellular inclusions stained by May-Grunwald Giemsa (MGG) were determined by spectrally resolved imaging. Multipixel spectra were obtained from Cabot rings and Howell-Jolly (HJ) bodies, displaying a range of wavelengths of transmitted light. The spectral characteristics of these inclusions were compared with those of isolated DNA, his-tones (type II) and arginine-rich histones (type VI), all stained by MGG. Results of single-cell spectroscopy show that the spectra of Cabot rings and HJ bodies share spectral characteristics with the type II and type VI histones. However, no resemblance was found between Cabot rings and DNA spectra. The spectral analysis of hetero-chromatin displayed a spectral pattern with characteristics of both DNA and histones, while the euchromatin showed a major contribution of the DNA component.     

Complex numbers in chemometrics: examples from multivariate impedance measurements on lipid monolayers

Electrical impedance gives multivariate complex number data as results. Two examples of multivariate electrical impedance data measured on lipid monolayers in different solutions give rise to matrices (16x50 and 38x50) of complex numbers. Multivariate data analysis by principal component analysis (PCA) or singular value decomposition (SVD) can be used for complex data and the necessary equations are given. The scores and loadings obtained are vectors of complex numbers. It is shown that the complex number PCA and SVD are better at concentrating information in a few components than the na?ve juxtaposition method and that Argand diagrams can replace score and loading plots. Different concentrations of Magainin and Gramicidin A give different responses and also the role of the electrolyte medium can be studied. An interaction of Gramicidin A in the solution with the monolayer over time can be observed.     

Subcellular Localization of Sulfonated Tetraphenyl Porphines in Colon Carcinoma Cells by Spectrally Resolved Imaging

Subcellular localization of the dye, 5,10,15,20-tetra(4-sul-fonatophenyl)porphine (TPPS₄) and the more hydrophobic dye, 5,10,15,20-tetra(1-sulfonatophenyl)porphine (TPPS₁), in murine colon carcinoma cells was studied by spectrally resolved imaging (SRI) combined with image processing techniques. Spectrally resolved imaging enabled the acquisition of multipixel fluorescence spectra (>10⁴) from a single cell. Demarcation of specific localization sites and segregation of the irrelevant fluorescence were based on the pixel spectra and by operating the functions of spectral similarity mapping (SSM), principal component analysis (PCA) and spectral classification. The SRI revealed the fine details of the photochemical process that clarify some aspects of subcellular damage. The SRI depicted the differences between TPPS₄ and TPPS, with respect to their initial localization and their fate at the end of the photochemical effect. The dye TPPS₄ was localized initially in lysosomal vesicles, and upon irradiation fluorescence was seen in the nucleus as well as in vesicles. Some of the vesicles were closely related to the nucleus, as resolved by SSM, PCA and spectral classification. Additional light exposure stimulated relocalization of TPPS₄ into the nucleus as well as into the nucleolus, which was clearly depicted by SSM and PCA. Spectral classification showed a third, weak residual cytoplasmic array around the nucleus. The dye TPPS, concentrated in a Golgi-like complex and was resolved in the nuclear envelope and in small vesicles: it was not redistributed into other compartments upon photosensitization. Serum supplementation to the incubation media of colon carcinoma cells treated with TPPS₄ or TPPS, did not change the localization patterns. Pixel spectra of the two dyes in the cells showed spectral shifts and expanded shoulders due to microenvironmental effects. Thus, the chemical nature of the sulfonated phenyl porphines, and not their interaction with serum proteins, was the main determinant of their binding to the lysosomes, nucleus, nucleolus, nuclear envelope or Golgi.     

PK-means: A new algorithm for gene clustering

Microarray technology has been widely applied in study of measuring gene expression levels for thousands of genes simultaneously. Gene cluster analysis is found useful for discovering the function of gene because co-expressed genes are likely to share the same biological function. K-means is one of well-known clustering methods. However, it is sensitive to the selection of an initial clustering and easily becoming trapped in a local minimum. Particle-pair optimizer (PPO) is a variation on the traditional particle swarm optimization (PSO) algorithm, which is stochastic particle-pair based optimization technique that can be applied to a wide range of problems. In this paper we bridges PPO and K-means within the algorithm PK-means for the first time. Our results indicate that PK-means clustering is generally more accurate than K-means and Fuzzy K-means (FKM). PK-means also has better robustness for it is less sensitive to the initial randomly selected cluster centroids. Finally, our algorithm outperforms these methods with fast convergence rate and low computation load.     

Cluster analysis of MS- and IR-spectra of pyrazines and related heterocyclic flavour substances

Summary The application of Principal Component Analysis (PCA) on MS- and IR-spectra of 77 substances from a test data set, belonging to 5 different classes of heterocyclic components (pyrazines, pyrroles, pyridines, thiazoles and quinoxalines, Table 1) resulted in a clear separation of the MS-spectral data in distinct clusters and led to the definition of planar classifiers, which were used for the detection of these classes of compounds in the spectral data set of a complex natural matrix. The projection of the GC-MS data of the headspace of opium in the plane of two main variances and the application of the planar classifier for pyrazines/pyrroles resulted in the reduction of the original data set by factor of 30 and allowed more efficient identification of 3 alkylpyrazines and 2 acylsubstituted pyrroles. The PCA of full dimensionality IR-spectra only resulted in less pronounced cluster separation.