Regulatory processes and history dependent behavior appear in many dynamical systems in nature and technology. For modeling regulatory processes, hybrid systems offer several advances. From this point of view, to observe the capability of hybrid systems in a history dependent system is a strong motivation. In this thesis, we developed functional hybrid systems which exhibit memory dependent behavior such that the dynamics of the system is determined by both the location of the state vector and the memory. This property was explained by various examples. We used the hybrid system with memory in modeling the gene regulatory network of human immune response to Influenza A virus infection. We investigated the sensitivity of the piecewise linear model with memory. We introduced how the model can be developed in future.

abstract: Synthetic gene networks have evolved from simple proof-of-concept circuits to complex therapy-oriented networks over the past fifteen years. This advancement has greatly facilitated expansion of the emerging field of synthetic biology. Multistability is a mechanism that cells use to achieve a discrete number of mutually exclusive states in response to environmental inputs. However, complex contextual connections of gene regulatory networks in natural settings often impede the experimental establishment of the function and dynamics of each specific gene network. In this work, diverse synthetic gene networks are rationally designed and constructed using well-characterized biological components to approach the cell fate determination and state transition dynamics in multistable systems. Results show that unimodality and bimodality and trimodality can be achieved through manipulation of the signal and promoter crosstalk in quorum-sensing systems, which enables bacterial cells to communicate with each other. Moreover, a synthetic quadrastable circuit is also built and experimentally demonstrated to have four stable steady states. Experiments, guided by mathematical modeling predictions, reveal that sequential inductions generate distinct cell fates by changing the landscape in sequence and hence navigating cells to different final states. Circuit function depends on the specific protein expression levels in the circuit. We then establish a protein expression predictor taking into account adjacent transcriptional regions’ features through construction of ~120 synthetic gene circuits (operons) in Escherichia coli. The predictor’s utility is further demonstrated in evaluating genes’ relative expression levels in construction of logic gates and tuning gene expressions and nonlinear dynamics of bistable gene networks. These combined results illustrate applications of synthetic gene networks to understand the cell fate determination and state transition dynamics in multistable systems. A protein-expression predictor is also developed to evaluate and tune circuit dynamics. / Dissertation/Thesis / Doctoral Dissertation Biomedical Engineering 2017

Arsenic is a universal environmental toxicant associated mostly with skin related diseases in people exposed to low doses over a long term. Low dose arsenic trioxide (ATO) with long exposure will lead to chronic exposure. Experiments were performed to provide new knowledge on the incompletely understood mechanisms of action of chronic low dose inorganic arsenic in keratinocytes. Cytotoxicity patterns of ATO on long-term cultures of HaCaT cells on collagen IV was studied over a time course of 14 days. DNA damage was also assessed. The percentages of viable cells after exposure were measured on Day 2, Day 5, Day 8, and Day 14. Statistical and visual analytics approaches were used for data analysis. In the result, a biphasic toxicity response was observed at a 5 μg/ml dose with cell viability peaking on Day 8 in both chronic and acute exposures. Furthermore, a low dose of 1 μg/ml ATO enhanced HaCaT keratinocyte proliferation but also caused DNA damage. Global gene expression study using microarray technique demonstrated differential expressions of genes in HaCaT cell exposed to 0.5 μg/ml dose of ATO up to 22 passages. Four of the up-regulated and 1 down-regulated genes were selected and confirmed with qRT-PCR technique. These include; Aldo-Keto Reductase family 1, member C3 (AKR1C3), Insulin Growth Factor-Like family member 1 (IGFL1), Interleukin 1 Receptor, type 2 (IL1R2) and Tumour Necrosis Factor [ligand] Super-Family, member 18 (TNFSF18), and down-regulated Regulator of G-protein Signalling 2 (RGS2). The decline in growth inhibiting gene (RGS2) and increase in AKR1C3 may be the contributory path to chronic inflammation leading to metaplasia. This pathway is proposed to be a mechanism leading to carcinogenesis in skin keratinocytes. The observed over expression of IGFL1 may be a means of triggering carcinogenesis in HaCaT keratinocytes. In conclusion, it was established that at very low doses, arsenic is genotoxic and induces aberrations in gene expression though it may appear to enhance cell proliferation. The expression of two genes encoding membrane proteins IL1R2 and TNFSF18 may serve as possible biomarkers of skin keratinocytes intoxication due to arsenic exposure. This research provides insights into previously unknown gene markers that may explain the mechanisms of arsenic-induced dermal disorders including skin cancer / Environmental Sciences / D. Phil. (Environmental science)

Alcohol Dependence (AD) is a chronic substance use disorder with moderate heritability (60%). Linkage and genome-wide association studies (GWAS) have implicated a number of loci; however, the molecular mechanisms underlying AD are unclear. Advances in systems biology allow genome-wide expression data to be integrated with genetic data to detect expression quantitative trait loci (eQTL), polymorphisms that regulate gene expression levels, influence phenotypes and are significantly enriched among validated genetic signals for many commonly studied traits including AD. We integrated genome-wide mRNA and miRNA expression data with genotypic data from the nucleus accumbens (NAc), a major addiction-related brain region, of 36 subjects (18 AD cases, 18 matched controls). We applied weighted gene co-expression network analysis (WGCNA) to identify mRNA and miRNA gene co-expression modules significantly associated with AD. We identified six mRNA modules, two of which were downregulated in AD and were enriched for neuronal marker gene expression. The remaining four modules were upregulated in AD and enriched for astrocyte and microglial marker gene expressions. After performing gene set enrichment analysis (GSEA), we found that neuronal-specific modules enriched for oxidative phosphorylation, mitochondrial dysfunction and MAPK signaling pathways and glial-specific modules enriched for immune related processes, cell adhesion molecules and cell signaling pathways. WGCNA was also applied to miRNA data and identified two downregulated and one upregulated modules in AD. We intersected computationally predicted miRNA:mRNA interactions with miRNA and mRNA expression correlations to identify 481 significant (FDR<0.10) miRNA:mRNA targeting pairs. Over half (54%) of the mRNAs were targeted cooperatively by more than one miRNA suggesting a potentially important cellular mechanism relevant to AD. After integrating our expression and genetic data we identified 591 significant mRNA and 68 significant miRNA cis-eQTLs (<1 megabase) (FDR<0.10). After querying against GWAS data from the Colaborative Study on Genetics of Alcohol and Study of Addiction: Gentics and Environment, eQTLs for neuronatin (NNAT; rs1780705), proteosome subunit type 5 (PSMB5; rs10137082), long non-coding RNA (PKI55; rs13392737), adaptor related protein complex 1 sigma one subunit (AP1S1; rs12079545) and translocation associate membrane protein 1 (TRAM1; rs13277972) were associated with AD or alcohol related phenotypes at p<10-4.

L’arrivée des technologies à haut-débit pour mesurer l’expression des gènes a permis l’utilisation de signatures génomiques pour prédire des conditions cliniques ou la survie du patient. Cependant de telles signatures ont des limitations, comme la dépendance au jeu de données d’entrainement et le manque de généralisation. Nous proposons un nouvel algorithme, Integration Transcriptome-Interactome (ITI) (Garcia et al.) pour extraire une signature generalisable prédisant la rechute métastatique dans le cancer du sein par superimposition d’un très large jeu de données d’interaction protèine-protèine sur de multiples jeux de données d’expression des gènes. Cette méthode ré-implemente l’algorithme Chuang et al. , avec la capacité supplémentaire d’extraire une signature génomique à partir de plusieurs jeux de donnés d’expression des gènes simultanément. Une analyse non-supervisée et une analyse supervisée ont été réalisés sur un compendium de jeux de donnés issus de puces à ADN en cancer du sein. Les performances des signatures trouvées par ITI ont été comparé aux performances des signatures préalablement publiées (Wang et al. , Van De Vijver et al. , Sotiriou et al. ). Nos résultats montrent que les signatures ITI sont plus stables et plus généralisables, et sont plus performantes pour classifier un jeu de données indépendant. Nous avons trouvés des sous-réseaux formant des complexes précédement relié à des fonctions biologiques impliquées dans la nétastase et le cancer du sein. Plusieurs gènes directeurs ont été détectés, dont CDK1, NCK1 et PDGFB, certains n’étant pas déjà relié à la rechute métastatique dans le cancer du sein. / High-throughput gene-expression profiling technologies yeild genomic signatures to predict clinical condition or patient outcome. However, such signatures have limitations, such as dependency on training set, and lack of generalization. We propose a novel algorithm, Interactome-Transcriptome Integration (ITI) (Garcia et al.) extract a generalizable signature predicting breast cancer relapse by superimposition of a large-scale protein-protein interaction data over several gene-expression data sets. This method re-implements the Chuang et al. algorithm, with the added capability to extract a genomic signature from several gene expression data sets simultaneously. A non-supervised and a supervised analysis were made with a breast cancer compendium of DNA microarray data sets. Performances of signatures found with ITI were compared with previously published signatures (Wang et al. , Van De Vijver et al. , Sotiriou et al. ). Our results show that ITI’s signatures are more stable and more generalizable, and perfom better when classifying an independant dataset. We found that subnetworks formed complexes functionally linked to biological functions related to metastasis and breast cancer. Several drivers genes were detected, including CDK1, NCK1 and PDGFB, some not previously linked to breast cancer relapse.

Depuis le premier séquençage du génome humain au début des années 2000, de grandes initiatives se sont lancé le défi de construire la carte des variabilités génétiques inter-individuelles, ou bien encore celle des altérations de l'ADN tumoral. Ces projets ont posé les fondations nécessaires à l'émergence de la médecine de précision, dont le but est d'intégrer aux dossiers médicaux conventionnels les spécificités génétiques d'un individu, afin de mieux adapter les traitements et les stratégies de prévention. La traduction des variations et des altérations de l'ADN en prédictions phénotypiques constitue toutefois un problème difficile. Les séquenceurs ou puces à ADN mesurent plus de variables qu'il n'y a d'échantillons, posant ainsi des problèmes statistiques. Les données brutes sont aussi sujettes aux biais techniques et au bruit inhérent à ces technologies. Enfin, les vastes réseaux d'interactions à l'échelle des protéines obscurcissent l'impact des variations génétiques sur le comportement de la cellule, et incitent au développement de modèles prédictifs capables de capturer un certain degré de complexité.Cette thèse présente de nouvelles contributions méthodologiques pour répondre à ces défis.Tout d'abord, nous définissons une nouvelle représentation des profils de mutations tumorales, qui exploite leur position dans les réseaux d'interaction protéine-protéine. Pour certains cancers, cette représentation permet d'améliorer les prédictions de survie à partir des données de mutations, et de stratifier les cohortes de patients en sous-groupes informatifs. Nous présentons ensuite une nouvelle méthode d'apprentissage permettant de gérer conjointement la normalisation des données et l'estimation d'un modèle linéaire. Nos expériences montrent que cette méthode améliore les performances prédictives par rapport à une gestion séquentielle de la normalisation puis de l'estimation. Pour finir, nous accélérons l'estimation de modèles linéaires parcimonieux, prenant en compte des interactions deux à deux, grâce à un nouvel algorithme. L'accélération obtenue rend cette estimation possible et efficace sur des jeux de données comportant plusieurs centaines de milliers de variables originales, permettant ainsi d'étendre la portée de ces modèles aux données des études d'associations pangénomiques. / Since the first sequencing of the human genome in the early 2000s, large endeavours have set out to map the genetic variability among individuals, or DNA alterations in cancer cells. They have laid foundations for the emergence of precision medicine, which aims at integrating the genetic specificities of an individual with its conventional medical record to adapt treatment, or prevention strategies.Translating DNA variations and alterations into phenotypic predictions is however a difficult problem. DNA sequencers and microarrays measure more variables than there are samples, which poses statistical issues. The data is also subject to technical biases and noise inherent in these technologies. Finally, the vast and intricate networks of interactions among proteins obscure the impact of DNA variations on the cell behaviour, prompting the need for predictive models that are able to capture a certain degree of complexity. This thesis presents novel methodological contributions to address these challenges. First, we define a novel representation for tumour mutation profiles that exploits prior knowledge on protein-protein interaction networks. For certain cancers, this representation allows improving survival predictions from mutation data as well as stratifying patients into meaningful subgroups. Second, we present a new learning framework to jointly handle data normalisation with the estimation of a linear model. Our experiments show that it improves prediction performances compared to handling these tasks sequentially. Finally, we propose a new algorithm to scale up sparse linear models estimation with two-way interactions. The obtained speed-up makes this estimation possible and efficient for datasets with hundreds of thousands of main effects, thereby extending the scope of such models to the data from genome-wide association studies.

A inferência de redes gênicas (GN) a partir de dados de expressão gênica temporal é um problema crucial e desafiador em Biologia Sistêmica. Os conjuntos de dados de expressão geralmente consistem em dezenas de amostras temporais e as redes consistem em milhares de genes, tornando inúmeros métodos de inferência inviáveis na prática. Para melhorar a escalabilidade dos métodos de inferência de GNs, esta tese propõe um arcabouço chamado GeNICE, baseado no modelo de redes gênicas probabilísticas. A principal novidade é a introdução de um procedimento de agrupamento de genes, com perfis de expressão relacionados, para fornecer uma solução aproximada com complexidade computacional reduzida. Os agrupamentos definidos são usados para reduzir a dimensionalidade permitindo uma busca exaustiva mais eficiente pelos melhores subconjuntos de genes preditores para cada gene alvo de acordo com funções critério multivariadas. GeNICE reduz consideravelmente o espaço de busca porque os candidatos a preditores ficam restritos a um gene representante por agrupamento. No final, uma análise multivariada é realizada para cada subconjunto preditor definido, visando recuperar subconjuntos mínimos para simplificar a rede gênica inferida. Em experimentos com conjuntos de dados sintéticos, GeNICE obteve uma redução substancial de tempo quando comparado a uma solução anterior sem a etapa de agrupamento, preservando a precisão da predição de expressão gênica mesmo quando o número de agrupamentos é pequeno (cerca de cinquenta) e o número de genes é grande (ordem de milhares). Para um conjunto de dados reais de microarrays de Plasmodium falciparum, a precisão da predição alcançada pelo GeNICE foi de aproximadamente 97% em média. As redes inferidas para os genes alvos da glicólise e do apicoplasto refletem propriedades topológicas de redes complexas do tipo \"mundo pequeno\" e \"livre de escala\", para os quais grande parte das conexões são estabelecidas entre os genes de um mesmo módulo e algumas poucas conexões fazem o papel de estabelecer uma ponte entre os módulos (redes mundo pequeno), e o grau de distribuição das conexões entre os genes segue uma lei de potência, na qual a maioria dos genes têm poucas conexões e poucos genes (hubs) apresentam um elevado número de conexões (redes livres de escala), como esperado. / Gene network (GN) inference from temporal gene expression data is a crucial and challenging problem in Systems Biology. Expression datasets usually consist of dozens of temporal samples, while networks consist of thousands of genes, thus rendering many inference methods unfeasible in practice. To improve the scalability of GN inference methods, this work proposes a framework called GeNICE, based on Probabilistic Gene Networks; the main novelty is the introduction of a clustering procedure to group genes with related expression profiles, to provide an approximate solution with reduced computational complexity. The defined clusters were used to perform an exhaustive search to retrieve the best predictor gene subsets for each target gene, according to multivariate criterion functions. GeNICE greatly reduces the search space because predictor candidates are restricted to one representative gene per cluster. Finally, a multivariate analysis is performed for each defined predictor subset to retrieve minimal subsets and to simplify the network. In experiments with in silico generated datasets, GeNICE achieved substantial computational time reduction when compared to an existing solution without the clustering step, while preserving the gene expression prediction accuracy even when the number of clusters is small (about fifty) relative to the number of genes (order of thousands). For a Plasmodium falciparum microarray dataset, the prediction accuracy achieved by GeNICE was roughly 97% on average. The inferred networks for the apicoplast and glycolytic target genes reflects the topological properties of \"small-world\"and \"scale-free\"complex network models in which a large part of the connections is established between genes of the same functional module (smallworld networks) and the degree distribution of the connections between genes tends to form a power law, in which most genes present few connections and few genes (hubs) present a large number of connections (scale-free networks), as expected.

A inferência de redes gênicas (GN) a partir de dados de expressão gênica temporal é um problema crucial e desafiador em Biologia Sistêmica. Os conjuntos de dados de expressão geralmente consistem em dezenas de amostras temporais e as redes consistem em milhares de genes, tornando inúmeros métodos de inferência inviáveis na prática. Para melhorar a escalabilidade dos métodos de inferência de GNs, esta tese propõe um arcabouço chamado GeNICE, baseado no modelo de redes gênicas probabilísticas. A principal novidade é a introdução de um procedimento de agrupamento de genes, com perfis de expressão relacionados, para fornecer uma solução aproximada com complexidade computacional reduzida. Os agrupamentos definidos são usados para reduzir a dimensionalidade permitindo uma busca exaustiva mais eficiente pelos melhores subconjuntos de genes preditores para cada gene alvo de acordo com funções critério multivariadas. GeNICE reduz consideravelmente o espaço de busca porque os candidatos a preditores ficam restritos a um gene representante por agrupamento. No final, uma análise multivariada é realizada para cada subconjunto preditor definido, visando recuperar subconjuntos mínimos para simplificar a rede gênica inferida. Em experimentos com conjuntos de dados sintéticos, GeNICE obteve uma redução substancial de tempo quando comparado a uma solução anterior sem a etapa de agrupamento, preservando a precisão da predição de expressão gênica mesmo quando o número de agrupamentos é pequeno (cerca de cinquenta) e o número de genes é grande (ordem de milhares). Para um conjunto de dados reais de microarrays de Plasmodium falciparum, a precisão da predição alcançada pelo GeNICE foi de aproximadamente 97% em média. As redes inferidas para os genes alvos da glicólise e do apicoplasto refletem propriedades topológicas de redes complexas do tipo \"mundo pequeno\" e \"livre de escala\", para os quais grande parte das conexões são estabelecidas entre os genes de um mesmo módulo e algumas poucas conexões fazem o papel de estabelecer uma ponte entre os módulos (redes mundo pequeno), e o grau de distribuição das conexões entre os genes segue uma lei de potência, na qual a maioria dos genes têm poucas conexões e poucos genes (hubs) apresentam um elevado número de conexões (redes livres de escala), como esperado. / Gene network (GN) inference from temporal gene expression data is a crucial and challenging problem in Systems Biology. Expression datasets usually consist of dozens of temporal samples, while networks consist of thousands of genes, thus rendering many inference methods unfeasible in practice. To improve the scalability of GN inference methods, this work proposes a framework called GeNICE, based on Probabilistic Gene Networks; the main novelty is the introduction of a clustering procedure to group genes with related expression profiles, to provide an approximate solution with reduced computational complexity. The defined clusters were used to perform an exhaustive search to retrieve the best predictor gene subsets for each target gene, according to multivariate criterion functions. GeNICE greatly reduces the search space because predictor candidates are restricted to one representative gene per cluster. Finally, a multivariate analysis is performed for each defined predictor subset to retrieve minimal subsets and to simplify the network. In experiments with in silico generated datasets, GeNICE achieved substantial computational time reduction when compared to an existing solution without the clustering step, while preserving the gene expression prediction accuracy even when the number of clusters is small (about fifty) relative to the number of genes (order of thousands). For a Plasmodium falciparum microarray dataset, the prediction accuracy achieved by GeNICE was roughly 97% on average. The inferred networks for the apicoplast and glycolytic target genes reflects the topological properties of \"small-world\"and \"scale-free\"complex network models in which a large part of the connections is established between genes of the same functional module (smallworld networks) and the degree distribution of the connections between genes tends to form a power law, in which most genes present few connections and few genes (hubs) present a large number of connections (scale-free networks), as expected.

<p>In every living cell, millions of different types of molecules constantly interact and react chemically in a complex system that can adapt to fluctuating environments and extreme conditions, living to survive and reproduce itself. The information required to produce these components is stored in the genome, which is copied in each cell division and transferred and mixed with another genome from parent to child. The regulatory mechanisms that control biological systems, for instance the regulation of expression levels for each gene, has evolved so that global robustness and ability to survive under harsh conditions is a strength, at the same time as biological tasks on a detailed molecular level must be carried out with good precision and without failures. This has resulted in systems that can be described as a hierarchy of levels of complexity: from the lowest level, where molecular mechanisms control other components at the same level, to pathways of coordinated interactions between components, formed to carry out particular biological tasks, and up to large-scale systems consisting of all components, connected in a network with a topology that makes the system robust and flexible. This thesis reports on work that model and analyze complex biological systems, and the signals and noise that regulate them, at all different levels of complexity. Also, it shows how signals are transduced vertically from one level to another, as when a single mutation can cause errors in low level mechanisms, disrupting pathways and create systemwide imbalances, such as in type 2 diabetes. The advancement of our knowledge of biological systems requires both that we go deeper and towards more detail, of single molecules in single cells, as well as taking a step back to understand the organisation and dynamics in the large networks of all components, and unite the different levels of complexity.</p>

In every living cell, millions of different types of molecules constantly interact and react chemically in a complex system that can adapt to fluctuating environments and extreme conditions, living to survive and reproduce itself. The information required to produce these components is stored in the genome, which is copied in each cell division and transferred and mixed with another genome from parent to child. The regulatory mechanisms that control biological systems, for instance the regulation of expression levels for each gene, has evolved so that global robustness and ability to survive under harsh conditions is a strength, at the same time as biological tasks on a detailed molecular level must be carried out with good precision and without failures. This has resulted in systems that can be described as a hierarchy of levels of complexity: from the lowest level, where molecular mechanisms control other components at the same level, to pathways of coordinated interactions between components, formed to carry out particular biological tasks, and up to large-scale systems consisting of all components, connected in a network with a topology that makes the system robust and flexible. This thesis reports on work that model and analyze complex biological systems, and the signals and noise that regulate them, at all different levels of complexity. Also, it shows how signals are transduced vertically from one level to another, as when a single mutation can cause errors in low level mechanisms, disrupting pathways and create systemwide imbalances, such as in type 2 diabetes. The advancement of our knowledge of biological systems requires both that we go deeper and towards more detail, of single molecules in single cells, as well as taking a step back to understand the organisation and dynamics in the large networks of all components, and unite the different levels of complexity.