Studies on immune activation in rheumatoid arthritis and reactive arthritis

Rheumatoid arthritis (RA) is an autoimmune disease characterized by synovitis, progressive joint destruction, and disability. Reactive arthritis (ReA) is a sterile joint inflammation following a distant mucosal infection. The clinical course of these diseases is variable and cannot be predicted with reasonable accuracy by clinical and laboratory markers. The predictive value of circulating soluble interleukin-2 receptor (sIL-2R), a marker of lymphocyte activation, measured by Immulite® automated immunoassay analyzer, was evaluated in two cohorts of RA patients. In 175 patients with active early RA randomized to treatment with either on disease-modifying antirheumatic drug (DMARD) or a combination of 3 DMARDs and prednisolone, low baseline sIL-2R level predicted remission after 6 months in patients treated with a single DMARD. In 24 patients with active RA refractory to DMARDs, low baseline sIL-2R level predicted rapid clinical response to treatment with infliximab, an anti-tumour necrosis factor antibody. Furthermore, in a cohort of 26 patients with acute ReA, high baseline sIL-2R level predicted remission after 6 months. Levels of circulating soluble E-selectin (sE-selectin), a marker of endothelial activation, were measured annually by enzyme-linked immunosorbent assay (ELISA) in a cohort of 85 patients with early RA. During a five-year follow-up, sE-selectin levels were associated with activity and outcome of RA. The levels of neutrophil and monocyte CD11b/CD18 expression measured by flow cytometry, and circulating levels of sE-selectin measured by ELISA, and procalcitonin by immunoluminometric assay, were compared in 28 patients with acute ReA and 16 patients with early RA. The levels of the markers were comparable in ReA, RA, and healthy control subjects. In conlusion, sIL-2R may provide a new predictive marker in early RA treated with a single DMARD and refractory RA treated with infliximab. In addition, sIL-2R level predicts remission in acute ReA.

Computing the Stochastic Complexity of Simple Probabilistic Graphical Models

Minimum Description Length (MDL) is an information-theoretic principle that can be used for model selection and other statistical inference tasks. There are various ways to use the principle in practice. One theoretically valid way is to use the normalized maximum likelihood (NML) criterion. Due to computational difficulties, this approach has not been used very often. This thesis presents efficient floating-point algorithms that make it possible to compute the NML for multinomial, Naive Bayes and Bayesian forest models. None of the presented algorithms rely on asymptotic analysis and with the first two model classes we also discuss how to compute exact rational number solutions.

A Probabilistic Approach to the Primary Visual Cortex

What can the statistical structure of natural images teach us about the human brain? Even though the visual cortex is one of the most studied parts of the brain, surprisingly little is known about how exactly images are processed to leave us with a coherent percept of the world around us, so we can recognize a friend or drive on a crowded street without any effort. By constructing probabilistic models of natural images, the goal of this thesis is to understand the structure of the stimulus that is the raison d etre for the visual system. Following the hypothesis that the optimal processing has to be matched to the structure of that stimulus, we attempt to derive computational principles, features that the visual system should compute, and properties that cells in the visual system should have. Starting from machine learning techniques such as principal component analysis and independent component analysis we construct a variety of sta- tistical models to discover structure in natural images that can be linked to receptive field properties of neurons in primary visual cortex such as simple and complex cells. We show that by representing images with phase invariant, complex cell-like units, a better statistical description of the vi- sual environment is obtained than with linear simple cell units, and that complex cell pooling can be learned by estimating both layers of a two-layer model of natural images. We investigate how a simplified model of the processing in the retina, where adaptation and contrast normalization take place, is connected to the nat- ural stimulus statistics. Analyzing the effect that retinal gain control has on later cortical processing, we propose a novel method to perform gain control in a data-driven way. Finally we show how models like those pre- sented here can be extended to capture whole visual scenes rather than just small image patches. By using a Markov random field approach we can model images of arbitrary size, while still being able to estimate the model parameters from the data.

Polymorphic low penetrance genes and breast cancer : The role of genes involved in metabolism of xenobiotics, estrogens and reactive oxygen species

Various endogenous and exogenous factors have been reported to increase the risk of breast cancer. Many of those are related to prolonged lifetime exposure to estrogens. Furthermore, a positive family history of breast cancer and certain benign breast diseases are known to increase the risk of breast cancer. The role of lifestyle factors, such as use of alcohol and smoking has been an area of intensive study. Alcohol has been found to increase the risk of breast cancer, whereas the role of smoking has remained obscure. A multitude of enzymes are involved in the metabolism of estrogens and xenobiotics including the carcinogens found in tobacco smoke. Many of the metabolic enzymes exhibit genetic polymorphisms that can lead to inter-individual differences in their abilities to modify hazardous substrates. Therefore, in presence of a given chemical exposure, one subgroup of women may be more susceptible to breast carcinogenesis, since they carry unfavourable forms of the polymorphic genes involved in the metabolism of the chemical. In this work, polymorphic genes encoding for cytochrome P450 (CYP) 1A1 and 1B1, N-acetyl transferase 2 (NAT2), sulfotransferase 1A1 (SULT1A1), manganese superoxide dismutase (MnSOD) and vitamin D receptor (VDR) were investigated in relation to breast cancer susceptibility in a Finnish population. CYP1A1, CYP1B1 and SULT1A1 are involved in the metabolism of both estrogens and xenobiotics, whereas NAT2 is involved only in the latter. MnSOD is an antioxidant enzyme protecting cells from oxidative damage. VDR, in turn, mediates the effects of the active form of vitamin D (1,25(OH)2D3, calcitriol) on maintenance of calcium homeostasis and it has anti-proliferative effects in many cancer cells. A 1.3-fold (95% CIs 1.01-1.73) increased risk of breast cancer was seen among women who carried the NAT2 slow acetylator genotype and a 1.5-fold (95% CI 1.1-2.0) risk was found in women with a MnSOD variant A allele containing genotypes compared to women with the NAT2 rapid acetylator genotype or to those with the MnSOD VV genotype, respectively. Instead, women with the VDR a allele containing genotypes were found to be at a decreased risk for breast cancer (OR 0.73; 95% CI 0.54-0.98) compared to women with the AA genotype. No significant overall associations were found between SULT1A1 or CYP genotypes and breast cancer risk, whereas a combination of the CYP1B1 432Val allele containing genotypes with the NAT2 slow acetylator genotypes posed a 1.5-fold (95% CI 1.03-2.24) increased risk. Moreover, NAT2 slow acetylator genotype was found to be confined to women with an advanced stage of breast cancer (stages III and IV). Further evidence for the association of xenobiotic metabolising genes with breast cancer risk was found when active smoking was taken into account. Women who smoked less than 10 cigarettes/day and carried at least one CYP1B1 432Val variant allele, were at 3.1-fold (95% CI 1.32-7.12) risk of breast cancer compared to women who smoked the same amount but did not carry the variant allele. Furthermore, the risk was significantly increased with increasing number of the CYP1B1 432Val alleles (p for trend 0.005). In addition, women who smoked less than 5 pack-years and carried the NAT2 slow acetylator genotype were at a 2.6-fold (95% CI 1.01-6.48) increased risk of breast cancer compared to women who smoked the same amount but carried the NAT2 rapid acetylator genotype. Furthermore, the combination of the CYP1B1 432Val allele and the NAT2 slow acetylator genotype increased the risk of breast cancer by 2.5-fold (95% CI 1.11-5.45) among ever smokers. Instead, the MnSOD A allele was found to be a risk factor among postmenopausal long-term smokers (>15 years of smoking) (OR 5.1; 95% CI 1.4-18.4) or among postmenopausal women who had smoked more than 10 cigarettes/day (OR 5.5; 95% CI 1.3-23.4) compared to women who had similar smoking habits but carried the MnSOD V/V genotype. Similarly, within subgroups of postmenopausal women who were using oral contraceptives, hormone replacement therapy or alcohol, women carrying the MnSOD A allele genotypes seemed to be at increased risk of breast cancer compared to women with the MnSOD V/V genotype. A positive family history of breast cancer and high parity were shown to be inversely associated with breast cancer risk among women carrying the VDR ApaI a allele or among premenopausal women carrying the SULT1A1*2 allele, respectively.

On cluster approximations of cellular automata

In this thesis I examine one commonly used class of methods for the analytic approximation of cellular automata, the so-called local cluster approximations. This class subsumes the well known mean-field and pair approximations, as well as higher order generalizations of these. While a straightforward method known as Bayesian extension exists for constructing cluster approximations of arbitrary order on one-dimensional lattices (and certain other cases), for higher-dimensional systems the construction of approximations beyond the pair level becomes more complicated due to the presence of loops. In this thesis I describe the one-dimensional construction as well as a number of approximations suggested for higher-dimensional lattices, comparing them against a number of consistency criteria that such approximations could be expected to satisfy. I also outline a general variational principle for constructing consistent cluster approximations of arbitrary order with minimal bias, and show that the one-dimensional construction indeed satisfies this principle. Finally, I apply this variational principle to derive a novel consistent expression for symmetric three cell cluster frequencies as estimated from pair frequencies, and use this expression to construct a quantitatively improved pair approximation of the well-known lattice contact process on a hexagonal lattice.

Probabilistic Inductive Querying Using ProbLog

We study how probabilistic reasoning and inductive querying can be combined within ProbLog, a recent probabilistic extension of Prolog. ProbLog can be regarded as a database system that supports both probabilistic and inductive reasoning through a variety of querying mechanisms. After a short introduction to ProbLog, we provide a survey of the different types of inductive queries that ProbLog supports, and show how it can be applied to the mining of large biological networks.

An Efficient Constraint Grammar Parser based on Inward Deterministic Automata

Pappret conceptualizes parsning med Constraint Grammar på ett nytt sätt som en process med två viktiga representationer. En representation innehåller lokala tvetydighet och den andra sammanfattar egenskaperna hos den lokala tvetydighet klasser. Båda representationer manipuleras med ren finite-state metoder, men deras samtrafik är en ad hoc -tillämpning av rationella potensserier. Den nya tolkningen av parsning systemet har flera praktiska fördelar, bland annat det inåt deterministiska sättet att beräkna, representera och räkna om alla potentiella tillämpningar av reglerna i meningen.

Building and Using Existing Hunspell Dictionaries and TEX Hyphenators as Finite-State Automata

There are numerous formats for writing spellcheckers for open-source systems and there are many descriptions for languages written in these formats. Similarly, for word hyphenation by computer there are TEX rules for many languages. In this paper we demonstrate a method for converting these spell-checking lexicons and hyphenation rule sets into finite-state automata, and present a new finite-state based system for writer’s tools used in current open-source software such as Firefox, OpenOffice.org and enchant via the spell-checking library voikko.