The future of Lake Baikal’s biodiversity is uncertain in response to climate change. Unlike its diverse benthos, Lake Baikal’s zooplankton is species poor, with up to 96% of its biomass being composed of a single Calanoid copepod species, Epischura baikalensis. This study characterizes the genetic differentiation and differential gene expression of E. baikalensis. Using partial-transcriptome sequences obtained by 454 Rosche and Illumina sequencing technologies, the genetic differentiation at inferred single nucleotide polymorphism (SNP) sites and differential gene expression in populations sampled from various parts of the lake were analyzed. The functional genomics of genes showed significant differential expression among the lake’s regions with some genes being highly up-or down-regulated. High genetic differentiation among regions suggests isolated subpopulations. Moreover, significantly differentially expressed transcripts were significantly more genetically differentiated than transcripts exhibiting no differential expression. These results suggest high potential phenotypic plasticity and adaptability in response to climate change, e.g., temperature.

La chute des prix des technologies de séquençage de nouvelle génération (NGS) s'est accompagnée de leur utilisation accrue, en recherche et en clinique. L'interprétation toujours meilleure des génomes humains peut permettre le développement de meilleures stratégies de prévention, de diagnostic et de traitement des maladies. Des investissements significatifs ont vu le jour dans de nombreux pays industrialisés en vue de réaliser les promesses de la médecine personnalisée. Cependant, le séquençage du génome complet de patients n'est offert en tant que test clinique que dans un nombre très limité d'établissements de santé dans le monde. La France et le Québec ont investi de manière considérable dans la recherche en génomique. Cependant, des décisions stratégiques doivent encore être prises quant à l'implémentation clinique des technologies NGS dans ces deux juridictions. Dès lors, l'objectif de ce projet est de contribuer à l'ensemble des preuves et faits à la disposition des décideurs publics. Nous avons focalisé notre attention sur deux technologies, le séquençage de l'exome (whole-exome sequencing, WES) et du génome complet (whole-genome sequencing, WGS). Notre objectif était d'établir si l'utilisation efficace et responsable du WES/WGS pouvait être mise en péril par des lacunes dans les politiques publiques ou cadres règlementaires et normatifs applicables. A l'heure actuelle, l'interprétation clinique de la séquence génomique ou exomique d'un patient nécessite l'intervention de nombreuses parties prenantes, y compris des chercheurs qui utilisent des outils, procédés et normes développés dans le cadre de la recherche pour analyser les données NGS. En parallèle, les cadres normatifs existants ont été construits pour accommoder les données génétiques, mais n'abordent pas la question des données génomiques. Notre hypothèse est que ces éléments créent un besoin de standardisation, qui pourrait requérir des adaptations du cadre normatif. Nous avons répondu à trois questions de recherches: (1) Quels enjeux les utilisateurs de technologies NGS soulèvent-il à propos de leur utilisation en clinique ? Pour répondre nous avons fait une étude systématique de la littérature. (2) Comment les données NGS de patients sont-elles à l'heure actuelle par des institutions de santé en France et au Québec ? Pour répondre nous avons réalisé une étude de cas multiples. (3) Y a-t-il des lacunes dans les cadres normatifs qui devraient être comblées pour assurer l'utilisation responsable, efficace et standardisée des données NGS en clinique ? [...] / The decreasing cost of next-generation sequencing (NGS) technologies has resulted in their increased use in research, and in the clinical context. Indeed, the correct interpretation of a human genome can enable better prevention, diagnosis and treatment strategies. Significant public investments in NGS have been made in various developed nations to realise the promise of personalized medicine. Yet, today the sequencing and analysis of a patient’s exome or genome is only offered as a clinical test in a limited number of clinics around the world. France and Quebec have made sizable investments in genomics research, and France announced the launch of a genomic medicine plan in 2016. However, policy decisions still have to be made on the nation-wide clinical implementation of NGS technologies in both jurisdictions. Therefore, this project’s objective was to contribute to the body of evidence available to policymakers in France and Quebec on the clinical implementation of NGS technologies. We focused our attention on two specific NGS technologies, namely Whole Genome Sequencing (WGS), and Whole Exome Sequencing (WES). We specifically aimed to assess if the responsible and efficient use of WES/WGS data in the context of clinical care could be impeded by policy gaps. Currently, the clinical interpretation of a patient’s genome sequence data is done through the intervention of many stakeholders including basic science researchers. These researchers use bioinformatics tools, processes and norms developed for research to filter and analyse patients NGS data. In parallel, existing regulatory and normative frameworks have been developed for the use of genetic data, and include no clear definition of genomic data or genomic technologies. We hypothesised that these elements create a strong need for standardization of practices, and may require adaptations of current regulatory and normative frameworks to the context of NGS. We therefore aimed to answer three research questions: (1) What issues do technology users experience and foresee when using WES data to inform patient care? To answer this, we performed a systematic review of the literature. (2) How are patients’ NGS data currently managed (produced, analysed, interpreted and shared) in clinical institutions in Quebec and in France? We answered this by performing a case studies analysis, interrogating key stakeholders directly involved in managing patients’ NGS data in France and Quebec. (3) Are there gaps in the current regulatory and normative frameworks which should be addressed to enable a responsible and efficient standardized use of NGS data in the clinic? [...]

början av 2000-talet sekvenserades det första hela mänskliga genomet och sedan dess har teknikerna förfinats, effektiviserats och förbättrats, så att utvecklingen på många områden, däribland forskningen om människans evolution, går fortare än någonsin tidigare. Från att ha sekvenserat en molekyl i taget är det idag möjligt att sekvensera miljontals molekyler parallellt och dessutom till ett så lågt pris att fler nu har möjlighet att använda sig av teknikerna. Bara under de senaste fem åren har det hänt mycket på området människans evolution och det är detta som vi kommer redogöra för i den här studien. Vi redovisar nya upptäckter –såsom att Homo naledi artbestämdes 2015 –, stärker gamla hypoteser –såsom att Homo sapiens utvandrade ur Afrika –samt stryker andra hypoteser –såsom att Australopithecus afarensis skulle vara den saknade länken mellan släktet Australopithecus och släktet Homo. Mycket har skett inom forskningen på människans evolution. Tillgång till nya biologiska tekniker har tillsammans med paleontologiska upptäckter bidragit till att skapa en större samstämmighet inom båda områden. Genom att teknikerna utvecklas och kompletterar varandra kan en större konsensus i resultaten skapas. Att utvecklingen av forskningen går fort fram har konsekvenser även för kurslitteraturen i gymnasieskolan. Vi analyserar därför om gymnasielitteraturen hinner med samt om förenklingarna som görs i den är rimlig. Det är svårt att avgöra om kurslitteraturen uppdateras i tillräcklig takt då förenklingarna är så pass stora. Vi anser att flera av de jämförda böckerna är tvetydiga i sina beskrivningar och framställer fakta som säkerställd, vilket inte alltid stämmer överens med hur området ser ut i verkligheten.

Gene activity is regulated at two separate layers. Through structural and chemical properties of DNA – the primary layer of encoding – local signatures may enable, or disable, the binding of proteins or complexes of them with regulatory potential to the DNA. At a higher level – the structural layer of encoding – gene activity is regulated through the properties of higher order DNA structure, chromatin, and chromosome organization. Cells with abnormal chromosome compaction or organization, e.g. cancer cells, may thus have perturbed regulatory activities resulting in abnormal gene activity. Hence, there is a great need to decode the transcriptional regulation encoded in both layers to further our understanding of the factors that control activity and life of a cell and, ultimately, an organism. Modern genome-wide studies with those aims rely on data-intense experiments requiring sophisticated computational and statistical methods for data handling and analyses. This thesis describes recent advances of analyzing experimental data from quantitative biological studies to decipher the structural layer of encoding in human cells. Adopting an integrative approach when possible, combining multiple sources of data, allowed us to study the influences of chromatin (Papers I and II) and chromosomal aberrations (Paper IV) on transcription. Combining chromatin data with chromosomal aberration data allowed us to identify putative driver oncogenes and tumor-suppressor genes in cancer (Paper IV). Bayesian approaches enabling the incorporation of background information in the models and the adaptability of such models to data have been very useful. Their usages yielded accurate and narrow detection of chromosomal breakpoints in cancer (Papers III and IV) and reliable positioning of nucleosomes and their dynamics during transcriptional regulation at functionally relevant regulatory elements (Paper II). Using massively parallel sequencing data, we explored the chromatin landscapes of human cells (Papers I and II) and concluded that there is a preferential and evolutionary conserved positioning at internal exons nearly unaffected by the transcriptional level. We also observed a strong association between certain histone modifications and the inclusion or exclusion of an exon in the mature gene transcript, suggesting a functional role in splicing.

Die Strahlentherapie ist neben der Chemotherapie und einer operativen Entfernung die stärkste Waffe für die Bekämpfung bösartiger Tumore in der Krebsmedizin. Nach Herz-Kreislauf-Erkrankungen ist Krebs die zweithäufigste Todesursache in der westlichen Welt, wobei Prostatakrebs heutzutage die häufigste, männliche Krebserkrankung darstellt. Trotz technologischer Fortschritte der radiologischen Verfahren kann es noch viele Jahre nach einer Radiotherapie zu einem Rezidiv kommen, was zum Teil auf die hohe Resistenzfähigkeit einzelner, entarteter Zellen des lokal vorkommenden Tumors zurückgeführt werden kann. Obwohl die moderne Strahlenbiologie viele Aspekte der Resistenzmechanismen näher beleuchtet hat, bleiben Fragestellungen, speziell über das zeitliche Ansprechen eines Tumors auf ionisierende Strahlung, größtenteils unbeantwortet, da systemweite Untersuchungen nur begrenzt vorliegen. Als Zellmodelle wurden vier Prostata-Krebszelllinien (PC3, DuCaP, DU-145, RWPE-1) mit unterschiedlichen Strahlungsempfindlichkeiten kultiviert und auf ihre Überlebensfähigkeit nach ionisierender Bestrahlung durch einen Trypanblau- und MTT-Vitalitätstest geprüft. Die proliferative Kapazität wurde mit einem Koloniebildungstest bestimmt. Die PC3 Zelllinie, als Strahlungsresistente, und die DuCaP Zelllinie, als Strahlungssensitive, zeigten dabei die größten Differenzen bezüglich der Strahlungsempfindlichkeit. Auf Grundlage dieser Ergebnisse wurden die beiden Zelllinien ausgewählt, um anhand ihrer transkriptomweiten Genexpressionen, eine Identifizierung potentieller Marker für die Prognose der Effizienz einer Strahlentherapie zu ermöglichen. Weiterhin wurde mit der PC3 Zelllinie ein Zeitreihenexperiment durchgeführt, wobei zu 8 verschiedenen Zeitpunkten nach Bestrahlung mit 1 Gy die mRNA mittels einer Hochdurchsatz-Sequenzierung quantifiziert wurde, um das dynamisch zeitversetzte Genexpressionsverhalten auf Resistenzmechanismen untersuchen zu können. Durch das Setzen eines Fold Change Grenzwertes in Verbindung mit einem P-Wert < 0,01 konnten aus 10.966 aktiven Genen 730 signifikant differentiell exprimierte Gene bestimmt werden, von denen 305 stärker in der PC3 und 425 stärker in der DuCaP Zelllinie exprimiert werden. Innerhalb dieser 730 Gene sind viele stressassoziierte Gene wiederzufinden, wie bspw. die beiden Transmembranproteingene CA9 und CA12. Durch Berechnung eines Netzwerk-Scores konnten aus den GO- und KEGG-Datenbanken interessante Kategorien und Netzwerke abgeleitet werden, wobei insbesondere die GO-Kategorien Aldehyd-Dehydrogenase [NAD(P)+] Aktivität (GO:0004030) und der KEGG-Stoffwechselweg der O-Glykan Biosynthese (hsa00512) als relevante Netzwerke auffällig wurden. Durch eine weitere Interaktionsanalyse konnten zwei vielversprechende Netzwerke mit den Transkriptionsfaktoren JUN und FOS als zentrale Elemente identifiziert werden. Zum besseren Verständnis des dynamisch zeitversetzten Ansprechens der strahlungsresistenten PC3 Zelllinie auf ionisierende Strahlung, konnten anhand der 10.840 exprimierten Gene und ihrer Expressionsprofile über 8 Zeitpunkte interessante Einblicke erzielt werden. Während es innerhalb von 30 min (00:00 - 00:30) nach Bestrahlung zu einer schnellen Runterregulierung der globalen Genexpression kommt, folgen in den drei darauffolgenden Zeitabschnitten (00:30 - 01:03; 01:03 - 02:12; 02:12 - 04:38) spezifische Expressionserhöhungen, die eine Aktivierung schützender Netzwerke, wie die Hochregulierung der DNA-Reparatursysteme oder die Arretierung des Zellzyklus, auslösen. In den abschließenden drei Zeitbereichen (04:38 - 09:43; 09:43 - 20:25; 20:25 - 42:35) liegt wiederum eine Ausgewogenheit zwischen Induzierung und Supprimierung vor, wobei die absoluten Genexpressionsveränderungen ansteigen. Beim Vergleich der Genexpressionen kurz vor der Bestrahlung mit dem letzten Zeitpunkt (00:00 - 42:53) liegen mit 2.670 die meisten verändert exprimierten Gene vor, was einer massiven, systemweiten Genexpressionsänderung entspricht. Signalwege wie die ATM-Regulierung des Zellzyklus und der Apoptose, des NRF2-Signalwegs nach oxidativer Stresseinwirkung und die DNA-Reparaturmechanismen der homologen Rekombination, des nicht-homologen End Joinings, der MisMatch-, der Basen-Exzision- und der Strang-Exzision-Reparatur spielen bei der zellulären Antwort eine tragende Rolle. Äußerst interessant sind weiterhin die hohen Aktivitäten RNA-gesteuerter Ereignisse, insbesondere von small nucleolar RNAs und Pseudouridin-Prozessen. Demnach scheinen diese RNA-modifizierenden Netzwerke einen bisher unbekannten funktionalen und schützenden Einfluss auf das Zellüberleben nach ionisierender Bestrahlung zu haben. All diese schützenden Netzwerke mit ihren zeitspezifischen Interaktionen sind essentiell für das Zellüberleben nach Einwirkung von oxidativem Stress und zeigen ein komplexes aber im Einklang befindliches Zusammenspiel vieler Einzelkomponenten zu einem systemweit ablaufenden Programm. / The use of radiotherapy in addition to chemotherapy and surgical removal is the most powerful instrument in the fight against malignant tumors in cancer medicine. After cardiovascular diseases, cancer is the second leading cause of death in the western world, in which prostate cancer is the most frequent male cancer. Despite continuous technological improvements in radiological instruments and prognosis, it may occur a recurrence up to many years after radiotherapy due to a high resistance capability of individual malignant cells of the locally occurring tumor. Although modern radiation biology has studied many aspects of the resistance mechanisms, questions are largely unanswered especially in regards to prognostic terms and time response of tumor cells to ionizing radiation. As cellular models four prostate cancer cell lines with different radiation sensitivities (PC3, DuCaP, DU-145, RWPE-1) were cultured and tested for their ability to survive after exposure to ionizing radiation by a trypane blue and MTT viability assay. The proliferative capacity of the four cell lines was determined using a colony formation assay. The PC3 cell line (radiation-resistant) and the DuCaP cell line (radiation-sensitive) showed the maximal differences in terms of radiation sensitivity. Based on these results the two cell lines were selected to allow identification of potential prognostic marker for predicting the effectiveness of radiation therapy via their transcriptome-wide gene expression. Furthermore, a time series experiment with the radiation-resistant PC3 cell line was performed. At 8 different time points, during the period from 00:00 - 42:53 (hh:mm) after exposure with 1 Gy, the mRNA was quantified by next generation sequencing to investigate the dynamic behavior of time-delayed gene expression and to discover resistance mechanisms. Of 10,966 expressed genes 730 were significant differentially expressed, determined by setting a fold change threshold in conjunction with a P-value < 0.01. Of those 305 were more strongly expressed in PC3 cell line and 425 were more strongly expressed in the DuCaP cell line. Within these 730 genes many known stress-associated genes could be found, such as the two trans-membrane protein genes CA9 and CA12, which are associated with increased radiation resistance. By calculating a network score interesting networks were derived by the GO and KEGG databases. In particular the GO categories aldehyde dehydrogenase [NAD(P)+] activity (GO:0004030) as well as the KEGG pathway of O-glycan biosynthesis (hsa00512) seems to be remarkably relevant. An interaction analysis revealed two promising networks with the transcription factors JUN and FOS as central elements. High expression of the JUN network would be stand as indicator for radiation resistance whereas a high expression of the FOS network is equated with radiation sensitivity. Interesting insights could be achieved by analyzing the 10,840 expressed genes of the PC3 cell line and its expression profile over the 8 time points. Shortly after irradiation (00:00 - 00:30) a transcriptome-wide down-regulation occurred, within the next three, short time periods (00:30 - 01:03; 01:03 - 02:12; 02:12 - 04:38) a predominant increase of gene expression and the activation of protective networks followed, such as the up-regulation of DNA repair systems or the arresting of cell cycle. In the ensuing three time periods (4:38 - 09:43; 09:43 - 20:25; 20:25 - 42:35) a balance between gene induction and suppression was present and the absolute gene expression change was increased. When comparing the gene expression prior to irradiation with the last time point (00:00 - 42:53) 2,670 genes were differentially expressed, suggesting a massive and system-wide change of gene expression. Signaling pathways such as the ATM-regulated cell cycle and apoptosis, the Nrf2 pathway after oxidative stress exposure, the DNA repair mechanisms of homologous recombination, the non-homologous end joining, the mismatch repair, base-excision repair and strand-excision repair play a major role. Very interesting are the high activity of RNA-driven events, especially activities of small nucleolar RNAs and pseudouridine processes. This suggests that these RNA-modifying networks could have a hitherto unknown functional and protective effect on cell survival after exposure to ionizing radiation. All these protective networks and their time-specific interactions are essential for the survival of cells after exposure to oxidative stress and show a complex but consistent interaction of many individual components to a system-wide running program.

La present Tesi Doctoral s’emmarca dins d’una línia d’investigació dedicada a l’estudi de les bases genètiques del metabolisme dels greixos en relació a la producció de carn de porcí d’alta qualitat i saludable. L’objectiu final és la identificació de polimorfismes i mecanismes de regulació responsables de la variabilitat genètica d’aquests caràcters complexes en l’espècie porcina. En aquest context, s’han utilitzat mètodes de genòmica estructural (mapes de QTL d’expressió (eQTL); estudis de gens candidats) i funcional (estudis d’expressió gènica) a més a més d’anàlisis proteics i cel·lulars en mostres de múscul i greix de porcs seleccionats per les seves característiques de qualitat de carn. Així, mitjançant la tècnica de microarrays s’ha analitzat el patró d’expressió d’ARNm en mostres de múscul gluteus medius obtingudes a partir de porcs d’una població comercial Duroc amb fenotips divergents per diversos paràmetres relacionats amb la deposició dels lípids. Com a resultat, s’han observat nombrosos gens diferencialment expressats entre els animals amb perfils divergents d’engreixament. Un estudi ontològic/funcional va revelar que aquests gens estaven particularment relacionats amb el metabolisme lipídic, el creixement i la diferenciació muscular, la immunitat i la captació de glucosa en la ruta de la insulina. D’altra banda, les anàlisis d’eQTL han revelat l’existència de regions genòmiques responsables de la variació de l’expressió gènica en el múscul gluteus medius porcí; algunes de les quals mostren una concordança posicional amb varis QTL per caràcters de qualitat de cran i engreixament detectats prèviament a la mateixa població Duroc. Complementàriament, s’ha realitzat un estudi més exhaustiu de quatre gens candidats (ACACA, HMGCR, SCD i 6D), directament implicats en caràcters relacionats amb la qualitat de la carn al ser els principals responsables de la síntesis i dessaturació d’àcids grassos i colesterol. Combinant els resultats de l’anàlisi d’expressió gènica, mapes d’eQTL i els gens candidats estudiats s’ha elaborat una llista de gens candidats funcionals i posicionals que serà la base de futures investigacions cap a l’establiment de les xarxes gèniques i els mecanismes moleculars implicats en el metabolisme dels lípids musculars i els caràcters relacionats amb la qualitat de la carn en porcí. / La presente Tesis Doctoral se enmarca en una línea de investigación dedicada al estudio de las bases genéticas del metabolismo de las grasas en relación a la producción de carne de porcino de alta calidad y saludable. El objetivo final es la identificación de polimorfismos y mecanismos de regulación responsables de la variabilidad genética de estos caracteres complejos en porcino. Para ello se han utilizado métodos de genómica estructural (mapas de QTL de expresión (eQTL); estudio de genes candidatos) y funcional (estudios de expresión génica) y también de análisis proteico y celular en muestras de músculo y grasa de cerdos seleccionados por sus características de calidad de carne. Así, mediante la técnica de microarrays se ha analizado el patrón de expresión de ARNm en muestras de músculo gluteus medius obtenidas a partir de cerdos de una población comercial Duroc con fenotipos divergentes para varios parámetros relacionados con la deposición de los lípidos. Como resultado, se han observado numerosos genes diferencialmente expresados entre los animales con perfiles divergentes de engorde. Un estudio ontológico/funcional mostró que estos genes estaban particularmente relacionados con el metabolismo lipídico, el crecimiento y la diferenciación muscular, la inmunidad, y la captación de glucosa en la ruta de la insulina. Por otra parte, un análisis de eQTL ha revelado la existencia de regiones genómicas responsables de la variación de la expresión génica en el músculo gluteus medius porcino, algunas de las cuales muestran una concordancia posicional con varios QTL para caracteres de calidad de carne y engorde detectados previamente en la misma población Duroc. Complementariamente, se ha realizado un estudio más exhaustivo de cuatro genes candidatos (ACACA, HMGCR, SCD y 6D) directamente implicados en caracteres relacionados con la calidad de la carne al ser los principales responsables de la síntesis y desaturación de ácidos grasos y colesterol. Combinando los resultados del análisis de expresión génica, mapas de eQTL y los genes candidatos estudiados se ha elaborado una lista de genes candidatos funcionales y posicionales que será la base de futuras investigaciones hacia el establecimiento de las redes génicas y los mecanismos moleculares implicados en el metabolismo de los lípidos musculares y los caracteres relacionados con la calidad de la carne en porcino. / This PhD is part of a line of research devoted to studying the genetic basis of lipid metabolism and fat deposition in pigs with a view to producing healthy and high quality meat. The main objective is the identification of polymorphisms and regulatory mechanisms responsible for the genetic variability of these complex characters in pigs. In this sense, we have used several methods in the fields of structural (expression QTL (eQTL) maps; candidate genes studies) and functional (gene expression studies) genomics and also protein and cell studies in muscle and fat samples from pigs selected by meat quality parameters. In this context, using microarrays we analyzed the mRNA expression pattern in gluteus medius muscle samples obtained from a commercial Duroc pig population with divergent phenotypes for several parameters related to lipid deposition. As a result, we have obtained a list of genes differentially expressed between animals with divergent profiles related to lipid deposition. The ontological/functional study showed that these genes were particularly related to lipid metabolism, growth and muscle differentiation, immunity and glucose uptake in the insulin pathway. Moreover, analysis of eQTL has revealed the existence of genomic regions responsible for the variation of gene expression in porcine gluteus medius muscle. Some of these eQTL show positional concordance with several QTL related to meat quality and fat deposition previously identified in the same Duroc population. Additionally, we have performed a comprehensive study of four candidate genes (ACACA, HMGCR, SCD and 6D) directly involved in traits related to meat quality, playing an important role in fatty acids and cholesterol synthesis and desaturation. Combining the results of gene expression analysis, eQTL maps and candidate genes studied have resulted in a list of functional and positional candidate genes representing a valuable contribution to the understanding of the genetic regulation of skeletal muscle individual gene expression in swine species. This is a first step towards disentangling gene networks and molecular mechanisms involved in muscular lipid metabolism and meat quality traits in pigs.

Gene-fusions are a prevalent class of genetic variants that are often employed as cancer biomarkers and therapeutic targets. In recent years, high-throughput sequencing of the cellular genome and transcriptome have emerged as a promising approach for the investigation of gene-fusions at the DNA and RNA level. Although, large volumes of sequencing data and complexity of gene-fusion structures presents unique computational challenges. This dissertation describes research that first addresses the bioinformatics challenges associated with the analysis of the massive volumes of sequencing data by developing bioinformatics pipeline and more applied integrated computational workflows. Application of high-throughput sequencing and the proposed bioinformatics approaches for the breast and ovarian cancer study reveals unexpected complex structures of gene-fusions and their functional significance in the onset and progression of cancer. Integrative analysis of gene-fusions at DNA and RNA level shows the key importance of the regulation of gene-fusion at the transcription level in cancer.

Recent Genome Wide Association Studies (GWAS) have identified a series of single nucleotide polymorphism (SNP)s that are associated with Alzheimer’s disease (AD). One of the SNPs, rs3851179 (G/A), is near the gene phosphatidylinositol-binding clathrin assembly protein (PICALM). To evaluate whether this SNP is associated with PICALM expression, we quantified PICALM mRNA in 56 brain cDNA samples. Using linear regression analysis, we analyzed PICALM expression relative to rs3851179, AD status, and cell type specific markers. An association was detected between rs3851179 and PICALM, microvessel mRNA, glial fibrillary acidic protein (GFAP) mRNA, and synaptophysin (SYN) mRNA. To gain clarity into other possible SNP mechanisms, we searched brain cDNA for PICALM splice variants. We identified several PICALM splice variants involving exons 13-19. To identify and gain an estimation of relative abundance of splice variants, we PCR-amplified across exons 13-20 in cDNA from six individuals, three rs3851179 GG individuals and three rs3851179 AA individuals. Sequencing the cloned isoforms we found that PICALM lacking exon 13 (delta 13) is the most abundant isoform. Other isoforms detected included deletion of exon 18-19. We targeted the latter part of the gene, exon 17-20, to investigate unequal allelic expression using next generation sequencing. Individuals heterozygous for rs76719109 (n= 35), located in exon 17, were used to study the abundance of G/T allele in cDNA and genomic DNA. When we analyzed the T:G allelic ratio, the variant lacking exons 18 and 19 showed unequal allelic expression (p-value < 0.001) in a subset of individuals. One individual was an outlier, showing overall unequal allelic expression, which maybe be harboring a rare mutation capable of modifying PICALM expression. The PICALM intronic SNP rs588076 was associated with delta 18-19 isoform splicing (p-value < 0.001). In conclusion, this study gained a greater insight into the role of AD genetics in PICALM expression and splicing.

To gain a better understanding of contaminating bacteria in bioethanol industry, we profiled the bacterial community structure in corn-based bioethanol fermentations and evaluated its correlation to environmental variables. Twenty-three batches of corn-mash sample were collected from six bioethanol facilities. The V4 region of the collective bacterial 16S rRNA genes was analyzed by Illumina Miseq sequencing to investigate the bacterial community structure. Non-metric multidimensional scaling (NMDS) ordination plots were constructed to visualize bacterial community structure groupings among different samples, as well as the effects of multiple environmental variables on community structure variation. Our results suggest that bacterial community structure is facility-specific, although there are two core bacterial phyla, Firmicutes and Proteobacteria. Feedstock, facility, and fermentation technology may explain the difference in community structure between different facilities. Lactic acid, the most important environmental variable that influences bacterial community structure grouping, could be utilized as an indicator of bacterial contamination. We also identified genes responsible for the multiple antibiotic-resistance phenotype of an Enterobacter cloacae strain isolated from a bioethanol fermentation facility. We performed PCR assays and revealed the presence of canonical genes encoding resistance to penicillin and erythromycin. However, a gene encoding resistance to virginiamycin was not detected.

As we gain increased understanding of genetic disorders and gene regulation more focus has turned towards complex interactions. Combinations of genes or gene and environmental factors have been suggested to explain the missing heritability behind complex diseases. Furthermore, gene activation and splicing seem to be governed by a complex machinery of histone modification (HM), transcription factor (TF), and DNA sequence signals. This thesis aimed to apply and develop multivariate machine learning methods for use on such biological problems. Monte Carlo feature selection was combined with rule-based classification to identify interactions between HMs and to study the interplay of factors with importance for asthma and allergy. Firstly, publicly available ChIP-seq data (Paper I) for 38 HMs was studied. We trained a classifier for predicting exon inclusion levels based on the HMs signals. We identified HMs important for splicing and illustrated that splicing could be predicted from the HM patterns. Next, we applied a similar methodology on data from two large birth cohorts describing asthma and allergy in children (Paper II). We identified genetic and environmental factors with importance for allergic diseases which confirmed earlier results and found candidate gene-gene and gene-environment interactions. In order to interpret and present the classifiers we developed Ciruvis, a web-based tool for network visualization of classification rules (Paper III). We applied Ciruvis on classifiers trained on both simulated and real data and compared our tool to another methodology for interaction detection using classification. Finally, we continued the earlier study on epigenetics by analyzing HM and TF signals in genes with or without evidence of bidirectional transcription (Paper IV). We identified several HMs and TFs with different signals between unidirectional and bidirectional genes. Among these, the CTCF TF was shown to have a well-positioned peak 60-80 bp upstream of the transcription start site in unidirectional genes.