Visualisation of bioinformatics datasets

Analysing the molecular polymorphism and interactions of DNA, RNA and proteins is of fundamental importance in biology. Predicting functions of polymorphic molecules is important in order to design more effective medicines. Analysing major histocompatibility complex (MHC) polymorphism is important for mate choice, epitope-based vaccine design and transplantation rejection etc. Most of the existing exploratory approaches cannot analyse these datasets because of the large number of molecules with a high number of descriptors per molecule. This thesis develops novel methods for data projection in order to explore high dimensional biological dataset by visualising them in a low-dimensional space. With increasing dimensionality, some existing data visualisation methods such as generative topographic mapping (GTM) become computationally intractable. We propose variants of these methods, where we use log-transformations at certain steps of expectation maximisation (EM) based parameter learning process, to make them tractable for high-dimensional datasets. We demonstrate these proposed variants both for synthetic and electrostatic potential dataset of MHC class-I. We also propose to extend a latent trait model (LTM), suitable for visualising high dimensional discrete data, to simultaneously estimate feature saliency as an integrated part of the parameter learning process of a visualisation model. This LTM variant not only gives better visualisation by modifying the project map based on feature relevance, but also helps users to assess the significance of each feature. Another problem which is not addressed much in the literature is the visualisation of mixed-type data. We propose to combine GTM and LTM in a principled way where appropriate noise models are used for each type of data in order to visualise mixed-type data in a single plot. We call this model a generalised GTM (GGTM). We also propose to extend GGTM model to estimate feature saliencies while training a visualisation model and this is called GGTM with feature saliency (GGTM-FS). We demonstrate effectiveness of these proposed models both for synthetic and real datasets. We evaluate visualisation quality using quality metrics such as distance distortion measure and rank based measures: trustworthiness, continuity, mean relative rank errors with respect to data space and latent space. In cases where the labels are known we also use quality metrics of KL divergence and nearest neighbour classifications error in order to determine the separation between classes. We demonstrate the efficacy of these proposed models both for synthetic and real biological datasets with a main focus on the MHC class-I dataset.

Expectation Propagation with Factorizing Distributions: A Gaussian Approximation and Performance Results for Simple Models

We discuss the expectation propagation (EP) algorithm for approximate Bayesian inference using a factorizing posterior approximation. For neural network models, we use a central limit theorem argument to make EP tractable when the number of parameters is large. For two types of models, we show that EP can achieve optimal generalization performance when data are drawn from a simple distribution.

Dependent component analysis: a hyperspectral unmixing algorithm

Linear unmixing decomposes a hyperspectral image into a collection of reflectance spectra of the materials present in the scene, called endmember signatures, and the corresponding abundance fractions at each pixel in a spatial area of interest. This paper introduces a new unmixing method, called Dependent Component Analysis (DECA), which overcomes the limitations of unmixing methods based on Independent Component Analysis (ICA) and on geometrical properties of hyperspectral data. DECA models the abundance fractions as mixtures of Dirichlet densities, thus enforcing the constraints on abundance fractions imposed by the acquisition process, namely non-negativity and constant sum. The mixing matrix is inferred by a generalized expectation-maximization (GEM) type algorithm. The performance of the method is illustrated using simulated and real data.

Three-dimensional ordered-subset expectation maximization iterative protocol for evaluation of left ventricular volumes and function by quantitative gated SPECT: a dynamic phantom study.

The purposes of this study were to characterize the performance of a 3-dimensional (3D) ordered-subset expectation maximization (OSEM) algorithm in the quantification of left ventricular (LV) function with (99m)Tc-labeled agent gated SPECT (G-SPECT), the QGS program, and a beating-heart phantom and to optimize the reconstruction parameters for clinical applications. METHODS: A G-SPECT image of a dynamic heart phantom simulating the beating left ventricle was acquired. The exact volumes of the phantom were known and were as follows: end-diastolic volume (EDV) of 112 mL, end-systolic volume (ESV) of 37 mL, and stroke volume (SV) of 75 mL; these volumes produced an LV ejection fraction (LVEF) of 67%. Tomographic reconstructions were obtained after 10-20 iterations (I) with 4, 8, and 16 subsets (S) at full width at half maximum (FWHM) gaussian postprocessing filter cutoff values of 8-15 mm. The QGS program was used for quantitative measurements. RESULTS: Measured values ranged from 72 to 92 mL for EDV, from 18 to 32 mL for ESV, and from 54 to 63 mL for SV, and the calculated LVEF ranged from 65% to 76%. Overall, the combination of 10 I, 8 S, and a cutoff filter value of 10 mm produced the most accurate results. The plot of the measures with respect to the expectation maximization-equivalent iterations (I x S product) revealed a bell-shaped curve for the LV volumes and a reverse distribution for the LVEF, with the best results in the intermediate range. In particular, FWHM cutoff values exceeding 10 mm affected the estimation of the LV volumes. CONCLUSION: The QGS program is able to correctly calculate the LVEF when used in association with an optimized 3D OSEM algorithm (8 S, 10 I, and FWHM of 10 mm) but underestimates the LV volumes. However, various combinations of technical parameters, including a limited range of I and S (80-160 expectation maximization-equivalent iterations) and low cutoff values (< or =10 mm) for the gaussian postprocessing filter, produced results with similar accuracies and without clinically relevant differences in the LV volumes and the estimated LVEF.

Optimisation du processus de développement du médicament grâce à la modélisation PK et les simulations d’études cliniques

Le développement d’un médicament est non seulement complexe mais les retours sur investissment ne sont pas toujours ceux voulus ou anticipés. Plusieurs médicaments échouent encore en Phase III même avec les progrès technologiques réalisés au niveau de plusieurs aspects du développement du médicament. Ceci se traduit en un nombre décroissant de médicaments qui sont commercialisés. Il faut donc améliorer le processus traditionnel de développement des médicaments afin de faciliter la disponibilité de nouveaux produits aux patients qui en ont besoin. Le but de cette recherche était d’explorer et de proposer des changements au processus de développement du médicament en utilisant les principes de la modélisation avancée et des simulations d’essais cliniques. Dans le premier volet de cette recherche, de nouveaux algorithmes disponibles dans le logiciel ADAPT 5® ont été comparés avec d’autres algorithmes déjà disponibles afin de déterminer leurs avantages et leurs faiblesses. Les deux nouveaux algorithmes vérifiés sont l’itératif à deux étapes (ITS) et le maximum de vraisemblance avec maximisation de l’espérance (MLEM). Les résultats de nos recherche ont démontré que MLEM était supérieur à ITS. La méthode MLEM était comparable à l’algorithme d’estimation conditionnelle de premier ordre (FOCE) disponible dans le logiciel NONMEM® avec moins de problèmes de rétrécissement pour les estimés de variances. Donc, ces nouveaux algorithmes ont été utilisés pour la recherche présentée dans cette thèse. Durant le processus de développement d’un médicament, afin que les paramètres pharmacocinétiques calculés de façon noncompartimentale soient adéquats, il faut que la demi-vie terminale soit bien établie. Des études pharmacocinétiques bien conçues et bien analysées sont essentielles durant le développement des médicaments surtout pour les soumissions de produits génériques et supergénériques (une formulation dont l'ingrédient actif est le même que celui du médicament de marque, mais dont le profil de libération du médicament est différent de celui-ci) car elles sont souvent les seules études essentielles nécessaires afin de décider si un produit peut être commercialisé ou non. Donc, le deuxième volet de la recherche visait à évaluer si les paramètres calculer d’une demi-vie obtenue à partir d'une durée d'échantillonnage réputée trop courte pour un individu pouvaient avoir une incidence sur les conclusions d’une étude de bioéquivalence et s’ils devaient être soustraits d’analyses statistiques. Les résultats ont démontré que les paramètres calculer d’une demi-vie obtenue à partir d'une durée d'échantillonnage réputée trop courte influençaient de façon négative les résultats si ceux-ci étaient maintenus dans l’analyse de variance. Donc, le paramètre de surface sous la courbe à l’infini pour ces sujets devrait être enlevé de l’analyse statistique et des directives à cet effet sont nécessaires a priori. Les études finales de pharmacocinétique nécessaires dans le cadre du développement d’un médicament devraient donc suivre cette recommandation afin que les bonnes décisions soient prises sur un produit. Ces informations ont été utilisées dans le cadre des simulations d’essais cliniques qui ont été réalisées durant la recherche présentée dans cette thèse afin de s’assurer d’obtenir les conclusions les plus probables. Dans le dernier volet de cette thèse, des simulations d’essais cliniques ont amélioré le processus du développement clinique d’un médicament. Les résultats d’une étude clinique pilote pour un supergénérique en voie de développement semblaient très encourageants. Cependant, certaines questions ont été soulevées par rapport aux résultats et il fallait déterminer si le produit test et référence seraient équivalents lors des études finales entreprises à jeun et en mangeant, et ce, après une dose unique et des doses répétées. Des simulations d’essais cliniques ont été entreprises pour résoudre certaines questions soulevées par l’étude pilote et ces simulations suggéraient que la nouvelle formulation ne rencontrerait pas les critères d’équivalence lors des études finales. Ces simulations ont aussi aidé à déterminer quelles modifications à la nouvelle formulation étaient nécessaires afin d’améliorer les chances de rencontrer les critères d’équivalence. Cette recherche a apporté des solutions afin d’améliorer différents aspects du processus du développement d’un médicament. Particulièrement, les simulations d’essais cliniques ont réduit le nombre d’études nécessaires pour le développement du supergénérique, le nombre de sujets exposés inutilement au médicament, et les coûts de développement. Enfin, elles nous ont permis d’établir de nouveaux critères d’exclusion pour des analyses statistiques de bioéquivalence. La recherche présentée dans cette thèse est de suggérer des améliorations au processus du développement d’un médicament en évaluant de nouveaux algorithmes pour des analyses compartimentales, en établissant des critères d’exclusion de paramètres pharmacocinétiques (PK) pour certaines analyses et en démontrant comment les simulations d’essais cliniques sont utiles.

On Convergence Properties of the EM Algorithm for Gaussian Mixtures

"Expectation-Maximization'' (EM) algorithm and gradient-based approaches for maximum likelihood learning of finite Gaussian mixtures. We show that the EM step in parameter space is obtained from the gradient via a projection matrix $P$, and we provide an explicit expression for the matrix. We then analyze the convergence of EM in terms of special properties of $P$ and provide new results analyzing the effect that $P$ has on the likelihood surface. Based on these mathematical results, we present a comparative discussion of the advantages and disadvantages of EM and other algorithms for the learning of Gaussian mixture models.

Hierarchical Mixtures of Experts and the EM Algorithm

We present a tree-structured architecture for supervised learning. The statistical model underlying the architecture is a hierarchical mixture model in which both the mixture coefficients and the mixture components are generalized linear models (GLIM's). Learning is treated as a maximum likelihood problem; in particular, we present an Expectation-Maximization (EM) algorithm for adjusting the parameters of the architecture. We also develop an on-line learning algorithm in which the parameters are updated incrementally. Comparative simulation results are presented in the robot dynamics domain.

Specialized genetic algorithm for transmission network expansion planning considering reliability

This paper presents a methodology to solve the transmission network expansion planning problem (TNEP) considering reliability and uncertainty in the demand. The proposed methodology provides an optimal expansion plan that allows the power system to operate adequately with an acceptable level of reliability and in an enviroment with uncertainness. The reliability criterion limits the expected value of the reliability index (LOLE - Loss Of Load Expectation) of the expanded system. The reliability is evaluated for the transmission system using an analytical technique based in enumeration. The mathematical model is solved, in a efficient way, using a specialized genetic algorithm of Chu-Beasley modified. Detailed results from an illustrative example are presented and discussed. © 2009 IEEE.

String-averaging expectation-maximization for maximum likelihood estimation in emission tomography

Fundação de Amparo à Pesquisa do Estado de São Paulo (FAPESP)

Expectation Conditional Maximization-Based Deformable Shape Registration

This paper addresses the issue of matching statistical and non-rigid shapes, and introduces an Expectation Conditional Maximization-based deformable shape registration (ECM-DSR) algorithm. Similar to previous works, we cast the statistical and non-rigid shape registration problem into a missing data framework and handle the unknown correspondences with Gaussian Mixture Models (GMM). The registration problem is then solved by fitting the GMM centroids to the data. But unlike previous works where equal isotropic covariances are used, our new algorithm uses heteroscedastic covariances whose values are iteratively estimated from the data. A previously introduced virtual observation concept is adopted here to simplify the estimation of the registration parameters. Based on this concept, we derive closed-form solutions to estimate parameters for statistical or non-rigid shape registrations in each iteration. Our experiments conducted on synthesized and real data demonstrate that the ECM-DSR algorithm has various advantages over existing algorithms.

Speeding up the EM algorithm for mixture model-based segmentation of magnetic resonance images

Mixture models implemented via the expectation-maximization (EM) algorithm are being increasingly used in a wide range of problems in pattern recognition such as image segmentation. However, the EM algorithm requires considerable computational time in its application to huge data sets such as a three-dimensional magnetic resonance (MR) image of over 10 million voxels. Recently, it was shown that a sparse, incremental version of the EM algorithm could improve its rate of convergence. In this paper, we show how this modified EM algorithm can be speeded up further by adopting a multiresolution kd-tree structure in performing the E-step. The proposed algorithm outperforms some other variants of the EM algorithm for segmenting MR images of the human brain. (C) 2004 Pattern Recognition Society. Published by Elsevier Ltd. All rights reserved.

Using the EM Algorithm to Train Neural Networks: Misconceptions and a New Algorithm for Multiclass Classification

The expectation-maximization (EM) algorithm has been of considerable interest in recent years as the basis for various algorithms in application areas of neural networks such as pattern recognition. However, there exists some misconceptions concerning its application to neural networks. In this paper, we clarify these misconceptions and consider how the EM algorithm can be adopted to train multilayer perceptron (MLP) and mixture of experts (ME) networks in applications to multiclass classification. We identify some situations where the application of the EM algorithm to train MLP networks may be of limited value and discuss some ways of handling the difficulties. For ME networks, it is reported in the literature that networks trained by the EM algorithm using iteratively reweighted least squares (IRLS) algorithm in the inner loop of the M-step, often performed poorly in multiclass classification. However, we found that the convergence of the IRLS algorithm is stable and that the log likelihood is monotonic increasing when a learning rate smaller than one is adopted. Also, we propose the use of an expectation-conditional maximization (ECM) algorithm to train ME networks. Its performance is demonstrated to be superior to the IRLS algorithm on some simulated and real data sets.

Use of the EM algorithm to detect QTL affecting multiple-traits in an across half-sib family analysis

QTL detection experiments in livestock species commonly use the half-sib design. Each male is mated to a number of females, each female producing a limited number of progeny. Analysis consists of attempting to detect associations between phenotype and genotype measured on the progeny. When family sizes are limiting experimenters may wish to incorporate as much information as possible into a single analysis. However, combining information across sires is problematic because of incomplete linkage disequilibrium between the markers and the QTL in the population. This study describes formulae for obtaining MLEs via the expectation maximization (EM) algorithm for use in a multiple-trait, multiple-family analysis. A model specifying a QTL with only two alleles, and a common within sire error variance is assumed. Compared to single-family analyses, power can be improved up to fourfold with multi-family analyses. The accuracy and precision of QTL location estimates are also substantially improved. With small family sizes, the multi-family, multi-trait analyses reduce substantially, but not totally remove, biases in QTL effect estimates. In situations where multiple QTL alleles are segregating the multi-family analysis will average out the effects of the different QTL alleles.

EM algorithm for GTM-FS

We propose a generative topographic mapping (GTM) based data visualization with simultaneous feature selection (GTM-FS) approach which not only provides a better visualization by modeling irrelevant features ("noise") using a separate shared distribution but also gives a saliency value for each feature which helps the user to assess their significance. This technical report presents a varient of the Expectation-Maximization (EM) algorithm for GTM-FS.

Using Inside-Outside Algorithm for Estimation of the Offspring Distribution in Multitype Branching Processes

Multitype branching processes (MTBP) model branching structures, where the nodes of the resulting tree are particles of different types. Usually such a process is not observable in the sense of the whole tree, but only as the “generation” at a given moment in time, which consists of the number of particles of every type. This requires an EM-type algorithm to obtain a maximum likelihood (ML) estimate of the parameters of the branching process. Using a version of the inside-outside algorithm for stochastic context-free grammars (SCFG), such an estimate could be obtained for the offspring distribution of the process.