Exploratory data analysis with non-linear and missing data in geochemistry

Exploratory analysis of data seeks to find common patterns to gain insights into the structure and distribution of the data. In geochemistry it is a valuable means to gain insights into the complicated processes making up a petroleum system. Typically linear visualisation methods like principal components analysis, linked plots, or brushing are used. These methods can not directly be employed when dealing with missing data and they struggle to capture global non-linear structures in the data, however they can do so locally. This thesis discusses a complementary approach based on a non-linear probabilistic model. The generative topographic mapping (GTM) enables the visualisation of the effects of very many variables on a single plot, which is able to incorporate more structure than a two dimensional principal components plot. The model can deal with uncertainty, missing data and allows for the exploration of the non-linear structure in the data. In this thesis a novel approach to initialise the GTM with arbitrary projections is developed. This makes it possible to combine GTM with algorithms like Isomap and fit complex non-linear structure like the Swiss-roll. Another novel extension is the incorporation of prior knowledge about the structure of the covariance matrix. This extension greatly enhances the modelling capabilities of the algorithm resulting in better fit to the data and better imputation capabilities for missing data. Additionally an extensive benchmark study of the missing data imputation capabilities of GTM is performed. Further a novel approach, based on missing data, will be introduced to benchmark the fit of probabilistic visualisation algorithms on unlabelled data. Finally the work is complemented by evaluating the algorithms on real-life datasets from geochemical projects.

Addressing missing data in geochemistry: a non-linear approach

Exploratory analysis of petroleum geochemical data seeks to find common patterns to help distinguish between different source rocks, oils and gases, and to explain their source, maturity and any intra-reservoir alteration. However, at the outset, one is typically faced with (a) a large matrix of samples, each with a range of molecular and isotopic properties, (b) a spatially and temporally unrepresentative sampling pattern, (c) noisy data and (d) often, a large number of missing values. This inhibits analysis using conventional statistical methods. Typically, visualisation methods like principal components analysis are used, but these methods are not easily able to deal with missing data nor can they capture non-linear structure in the data. One approach to discovering complex, non-linear structure in the data is through the use of linked plots, or brushing, while ignoring the missing data. In this paper we introduce a complementary approach based on a non-linear probabilistic model. Generative topographic mapping enables the visualisation of the effects of very many variables on a single plot, while also dealing with missing data. We show how using generative topographic mapping also provides an optimal method with which to replace missing values in two geochemical datasets, particularly where a large proportion of the data is missing.

TDP-43-Immunoreactive pathology in frontotemporal lobar degeneration with TDP proteinopathy (FTLD-TDP) with and without associated motor neuron disease

A proportion of patients with motor neuron disease (MND) exhibit frontotemporal dementia (FTD) and some patients with FTD develop the clinical features of MND. Frontotemporal lobar degeneration (FTLD) is the pathological substrate of FTD and some forms of this disease (referred to as FTLD-U) share with MND the common feature of ubiquitin-immunoreactive, tau-negative cellular inclusions in the cerebral cortex and hippocampus. Recently, the transactive response (TAR) DNA-binding protein of 43 kDa (TDP-43) has been found to be a major protein of the inclusions of FTLD-U with or without MND and these cases are referred to as FTLD with TDP-43 proteinopathy (FTLD-TDP). To clarify the relationship between MND and FTLD-TDP, TDP-43 pathology was studied in nine cases of FTLD-MND and compared with cases of familial and sporadic FTLD–TDP without associated MND. A principal components analysis (PCA) of the nine FTLD-MND cases suggested that variations in the density of surviving neurons in the frontal cortex and neuronal cytoplasmic inclusions (NCI) in the dentate gyrus (DG) were the major histological differences between cases. The density of surviving neurons in FTLD-MND was significantly less than in FTLD-TDP cases without MND, and there were greater densities of NCI but fewer neuronal intranuclear inclusions (NII) in some brain regions in FTLD-MND. A PCA of all FTLD-TDP cases, based on TDP-43 pathology alone, suggested that neuropathological heterogeneity was essentially continuously distributed. The FTLD-MND cases exhibited consistently high loadings on PC2 and overlapped with subtypes 2 and 3 of FTLD-TDP. The data suggest: (1) FTLD-MND cases have a consistent pathology, variations in the density of NCI in the DG being the major TDP-43-immunoreactive difference between cases, (2) there are considerable similarities in the neuropathology of FTLD-TDP with and without MND, but with greater neuronal loss in FTLD-MND, and (3) FTLD-MND cases are part of the FTLD-TDP ‘continuum’ overlapping with FTLD-TDP disease subtypes 2 and 3.

Quantitative neuropathology:data collection and statistical analysis

The use of quantitative methods has become increasingly important in the study of neuropathology and especially in neurodegenerative disease. Disorders such as Alzheimer's disease (AD) and the frontotemporal dementias (FTD) are characterized by the formation of discrete, microscopic, pathological lesions which play an important role in pathological diagnosis. This chapter reviews the advantages and limitations of the different methods of quantifying pathological lesions in histological sections including estimates of density, frequency, coverage, and the use of semi-quantitative scores. The sampling strategies by which these quantitative measures can be obtained from histological sections, including plot or quadrat sampling, transect sampling, and point-quarter sampling, are described. In addition, data analysis methods commonly used to analysis quantitative data in neuropathology, including analysis of variance (ANOVA), polynomial curve fitting, multiple regression, classification trees, and principal components analysis (PCA), are discussed. These methods are illustrated with reference to quantitative studies of a variety of neurodegenerative disorders.

A quantitative study of α-synuclein pathology in fifteen cases of dementia associated with Parkinson disease

The α-synuclein-immunoreactive pathology of dementia associated with Parkinson disease (DPD) comprises Lewy bodies (LB), Lewy neurites (LN), and Lewy grains (LG). The densities of LB, LN, LG together with vacuoles, neurons, abnormally enlarged neurons (EN), and glial cell nuclei were measured in fifteen cases of DPD. Densities of LN and LG were up to 19 and 70 times those of LB, respectively, depending on region. Densities were significantly greater in amygdala, entorhinal cortex (EC), and sectors CA2/CA3 of the hippocampus, whereas middle frontal gyrus, sector CA1, and dentate gyrus were least affected. Low densities of vacuoles and EN were recorded in most regions. There were differences in the numerical density of neurons between regions, but no statistical difference between patients and controls. In the cortex, the density of LB and vacuoles was similar in upper and lower laminae, while the densities of LN and LG were greater in upper cortex. The densities of LB, LN, and LG were positively correlated. Principal components analysis suggested that DPD cases were heterogeneous with pathology primarily affecting either hippocampus or cortex. The data suggest in DPD: (1) ratio of LN and LG to LB varies between regions, (2) low densities of vacuoles and EN are present in most brain regions, (3) degeneration occurs across cortical laminae, upper laminae being particularly affected, (4) LB, LN and LG may represent degeneration of the same neurons, and (5) disease heterogeneity may result from variation in anatomical pathway affected by cell-to-cell transfer of α-synuclein. © 2013 Springer-Verlag Wien.

Properties of intention:component structure and consequences for behavior, information processing and resistance

Three studies tested the impact of properties of behavioral intention on intention-behavior consistency, information processing, and resistance. Principal components analysis showed that properties of intention formed distinct factors. Study 1 demonstrated that temporal stability, but not the other intention attributes, moderated intention-behavior consistency. Study 2 found that greater stability of intention was associated with improved memory performance. In Study 3, participants were confronted with a rating scale manipulation designed to alter their intention scores. Findings showed that stable intentions were able to withstand attack. Overall, the present research findings suggest that different properties of intention are not simply manifestations of a single underlying construct ("intention strength"), and that temporal stability exhibits superior resistance and impact compared to other intention attributes. © 2013 Wiley Periodicals, Inc.

The classification of HLA supertypes by GRID/CPCA and hierarchical clustering methods

Biological experiments often produce enormous amount of data, which are usually analyzed by data clustering. Cluster analysis refers to statistical methods that are used to assign data with similar properties into several smaller, more meaningful groups. Two commonly used clustering techniques are introduced in the following section: principal component analysis (PCA) and hierarchical clustering. PCA calculates the variance between variables and groups them into a few uncorrelated groups or principal components (PCs) that are orthogonal to each other. Hierarchical clustering is carried out by separating data into many clusters and merging similar clusters together. Here, we use an example of human leukocyte antigen (HLA) supertype classification to demonstrate the usage of the two methods. Two programs, Generating Optimal Linear Partial Least Square Estimations (GOLPE) and Sybyl, are used for PCA and hierarchical clustering, respectively. However, the reader should bear in mind that the methods have been incorporated into other software as well, such as SIMCA, statistiXL, and R.

Forecasting the term structure when short-term rates are near zero

This paper compares the experience of forecasting the UK government bond yield curve before and after the dramatic lowering of short-term interest rates from October 2008. Out-of-sample forecasts for 1, 6 and 12 months are generated from each of a dynamic Nelson-Siegel model, autoregressive models for both yields and the principal components extracted from those yields, a slope regression and a random walk model. At short forecasting horizons, there is little difference in the performance of the models both prior to and after 2008. However, for medium- to longer-term horizons, the slope regression provided the best forecasts prior to 2008, while the recent experience of near-zero short interest rates coincides with a period of forecasting superiority for the autoregressive and dynamic Nelson-Siegel models. © 2014 John Wiley & Sons, Ltd.

Singapore's online retail deviants:analyzing the rise of blogshops' power

Geography, retailing, and power are institutionally bound up together. Within these, the authors situate their research in Clegg's work on power. Online shopping offers a growing challenge to the apparent hegemony of traditional physical retail stores' format. While novel e-formats appear regularly, blogshops in Singapore are enjoying astonishing success that has taken the large retailers by surprise. Even though there are well-developed theoretical frameworks for understanding the role of institutional entrepreneurs and other major stakeholders in bringing about change and innovation, much less attention has been paid to the role of unorganized, nonstrategic actors-such as blogshops-in catalyzing retail change. The authors explore how blogshops are perceived by consumers and how they challenge the power of other shopping formats. They use Principal Components Analysis to analyze results from a survey of 349 blogshops users. While the results show that blogshops stay true to traditional online shopping attributes, deviations occur on the concept of value. Furthermore, consumer power is counter intuitively found to be strongly present in the areas related to cultural ties, excitement, and search for individualist novelty (as opposed to mass-production), thereby encouraging researchers to think critically about emerging power behavior in media practices.

Comparative quantitative study of ‘signature’ pathological lesions in the hippocampus and adjacent gyri of 12 neurodegenerative disorders

The hippocampus (HC) and adjacent gyri are implicated in dementia in several neurodegenerative disorders. To compare HC pathology among disorders, densities of ‘signature’ pathological lesions were measured at a standard location in eight brain regions of 12 disorders. Principal components analysis of the data suggested that the disorders could be divided into three groups: (1) Alzheimer’s disease (AD), Down’s syndrome (DS), sporadic Creutzfeldt–Jakob disease, and variant Creutzfeldt–Jakob disease in which either β-amyloid (Aβ) or prion protein deposits were distributed in all sectors of the HC and adjacent gyri, with high densities being recorded in the parahippocampal gyrus and subiculum; (2) Pick’s disease, sporadic frontotemporal lobar degeneration with TDP-43 immunoreactive inclusions, and neuronal intermediate filament inclusion disease in which relatively high densities of neuronal cytoplasmic inclusions were present in the dentate gyrus (DG) granule cells; and (3) Parkinson’s disease dementia, dementia with Lewy bodies, progressive supranuclear palsy, corticobasal degeneration, and multiple system atrophy in which densities of signature lesions were relatively low. Variation in density of signature lesions in DG granule cells and CA1 were the most important sources of neuropathological variation among disorders. Hence, HC and adjacent gyri are differentially affected in dementia reflecting either variation in vulnerability of hippocampal neurons to specific molecular pathologies or in the spread of pathological proteins to the HC. Information regarding the distribution of pathology could ultimately help to explain variations in different cognitive domains, such as memory, observed in various disorders.

Quantitative pathological changes in the cerebellum of multiple system atrophy

Multiple system atrophy (MSA) is a rare neurodegenerative disorder associated with parkinsonism, ataxia, and autonomic dysfunction. Its pathology is primarily subcortical comprising vacuolation, neuronal loss, gliosis, and α-synuclein-immunoreactive glial cytoplasmic inclusions (GO). To quantify cerebellar pathology in MSA, the density and spatial pattern of the pathological changes were studied in α-synuclein-immunolabelled sections of the cerebellar hemisphere in 10 MSA and 10 control cases. In MSA, densities of Purkinje cells (PC) were decreased and vacuoles in the granule cell layer (GL) increased compared with controls. In six MSA cases, GCI were present in cerebellar white matter. In the molecular layer (ML) and GL of MSA, vacuoles were clustered, the clusters exhibiting a regular distribution parallel to the edge of the folia. Purkinje cells were randomly or regularly distributed with large gaps between surviving cells. Densities of glial cells and surviving neurons in the ML and surviving cells and vacuoles in the GL were negatively correlated consistent with gliosis and vacuolation in response to neuronal loss. Principal components analysis (PCA) suggested vacuole densities in the ML and vacuole density and cell losses in the GL were the main source of neuropathological variation among cases. The data suggest that: (1) cell losses and vacuolation of the GCL and loss of PC were the most significant pathological changes in the cases studied, (2) pathological changes were topographically distributed, and (3) cerebellar pathology could influence cerebral function in MSA via the cerebello-dentato-thalamic tract.

TDP-43-immunoreactive pathology in frontotemporal lobar degeneration with TDP proteinopathy (FTLD-TDP) with and without associated motor neuron disease (MND)

A proportion of patients with motor neuron disease (MND) exhibit frontotemporal dementia (FTD) and some patients with FTD develop the clinical features of MND. Frontotemporal lobar degeneration (FTLD) is the pathological substrate of FTD and some forms of this disease (referred to as FTLD-U) share with MND the common feature of ubiquitin-immunoreactive, tau-negative cellular inclusions in the cerebral cortex and hippocampus. Recently, the transactive response (TAR) DNA-binding protein of 43 kDa (TDP-43) has been found to be a major protein of the inclusions of FTLD-U with or without MND and these cases are referred to as FTLD with TDP-43 proteinopathy (FTLD-TDP). To clarify the relationship between MND and FTLD-TDP, TDP-43 pathology was studied in nine cases of FTLD-MND and compared with cases of familial and sporadic FTLD-TDP without associated MND. A principal components analysis (PCA) of the nine FTLD-MND cases suggested that variations in the density of surviving neurons in the frontal cortex and neuronal cytoplasmic inclusions (NCI) in the dentate gyrus (DG) were the major histological differences between cases. The density of surviving neurons in FTLD-MND was significantly less than in FTLD-TDP cases without MND, and there were greater densities of NCI but fewer neuronal intranuclear inclusions (NII) in some brain regions in FTLD-MND. A PCA of all FTLD-TDP cases, based on TDP-43 pathology alone, suggested that neuropathological heterogeneity was essentially continuously distributed. The FTLD-MND cases exhibited consistently high loadings on PC2 and overlapped with subtypes 2 and 3 of FTLD-TDP. The data suggest: (1) FTLD-MND cases have a consistent pathology, variations in the density of NCI in the DG being the major TDP-43-immunoreactive difference between cases, (2) there are considerable similarities in the neuropathology of FTLD-TDP with and without MND, but with greater neuronal loss in FTLD-MND, and (3) FTLD-MND cases are part of the FTLD-TDP 'continuum' overlapping with FTLD-TDP disease subtypes 2 and 3. © 2012 Nova Science Publishers, Inc. All rights reserved.

A quantitative study of tau pathology in eleven cases of chronic traumatic encephalopathy

AIMS: To quantify tau pathology of chronic traumatic encephalopathy (CTE) and investigate influence of dot-like lesions (DL), brain region, co-morbidity, and sporting career length. METHODS: Densities of neurofibrillary tangles (NFT), astrocytic tangles (AT), DL, oligodendroglial inclusions (GI), neuropil threads (NT), vacuoles, neurons, and enlarged neurons (EN) were measured in tau-immunoreactive sections of upper cortical laminae of frontal and temporal lobe, hippocampus (HC), amygdala, and substantia nigra (SN) of eleven cases of CTE. RESULTS: DL were a consistent finding in CTE. Densities of NFT, NT and DL were greatest in sectors CA1 and CA2 of the HC. Densities of AT were lower than NFT, small numbers of GI were recorded in temporal lobe, and low densities of vacuoles and EN were consistently present. β-amyloid containing neuritic plaques (NP) also occurred at low density. Densities of NFT, NT, DL, and AT were greater in sulci than gyri while vacuole density was greater in gyri. Principal components analysis (PCA) suggested that sporting career length and densities of NFT in entorhinal cortex, NT in CA2 and SN, and vacuolation in the DG were significant sources of variation among cases. CONCLUSION: DL are frequent in CTE suggesting affinity with argyrophilic grain disease (AGD) and Parkinson's disease dementia (PD-Dem). Densities of AT in all regions and NT/DL in sectors CA2/4 were consistent features of CTE. The eleven cases are neuropathologically heterogeneous which may result from genetic diversity, and variation in anatomical pathways subjected to trauma. This article is protected by copyright. All rights reserved.

Facial affect processing inbipolar disorder:a magnetoencephalography study

Background: Identifying biological markers to aid diagnosis of bipolar disorder (BD) is critically important. To be considered a possible biological marker, neural patterns in BD should be discriminant from those in healthy individuals (HI). We examined patterns of neuromagnetic responses revealed by magnetoencephalography (MEG) during implicit emotion-processing using emotional (happy, fearful, sad) and neutral facial expressions, in sixteen BD and sixteen age- and gender-matched healthy individuals. Methods: Neuromagnetic data were recorded using a 306-channel whole-head MEG ELEKTA Neuromag System, and preprocessed using Signal Space Separation as implemented in MaxFilter (ELEKTA). Custom Matlab programs removed EOG and ECG signals from filtered MEG data, and computed means of epoched data (0-250ms, 250-500ms, 500-750ms). A generalized linear model with three factors (individual, emotion intensity and time) compared BD and HI. A principal component analysis of normalized mean channel data in selected brain regions identified principal components that explained 95% of data variation. These components were used in a quadratic support vector machine (SVM) pattern classifier. SVM classifier performance was assessed using the leave-one-out approach. Results: BD and HI showed significantly different patterns of activation for 0-250ms within both left occipital and temporal regions, specifically for neutral facial expressions. PCA analysis revealed significant differences between BD and HI for mild fearful, happy, and sad facial expressions within 250-500ms. SVM quadratic classifier showed greatest accuracy (84%) and sensitivity (92%) for neutral faces, in left occipital regions within 500-750ms. Conclusions: MEG responses may be used in the search for disease specific neural markers.

Chemometric modelling to relate antioxidants, neutral lipid fatty acids and flavour components in chicken breast

Relationships among quality factors in retailed free-range, corn-fed, organic, and conventional chicken breasts (9) were modeled using chemometric approaches. Use of principal component analysis (PCA) to neutral lipid composition data explained the majority (93%) of variability (variance) in fatty acid contents in 2 significant multivariate factors. PCA explained 88 and 75% variance in 3 factors for, respectively, flame ionization detection (FID) and nitrogen phosphorus (NPD) components in chromatographic flavor data from cooked chicken after simultaneous distillation extraction. Relationships to tissue antioxidant contents were modeled. Partial least square regression (PLS2), interrelating total data matrices, provided no useful models. By using single antioxidants as Y variables in PLS (1), good models (r2 values > 0.9) were obtained for alpha-tocopherol, glutathione, catalase, glutathione peroxidase, and reductase and FID flavor components and among the variables total mono and polyunsaturated fatty acids and subsets of FID, and saturated fatty acid and NPD components. Alpha-tocopherol had a modest (r2 = 0.63) relationship with neutral lipid n-3 fatty acid content. Such factors thus relate to flavor development and quality in chicken breast meat.