Search

Amyotrophic Lateral Sclerosis (ALS) is a devastating neurodegenerative disease characterized by the progressive degeneration of upper and lower motor neurons. The progression of the disease is variable, with median survival of three years post disease onset. Riluzole, approved in 1995, is the only drug globally licenced for the treatment of ALS. However, riluzole has limited efficacy, prolonging patients' life span by only 3 months. New therapies are therefore urgently needed. This work sought to (i) utilize computational systems biology tools to investigate patient-phenotype heterogeneity and identify novel therapeutic targets and disease pathways in ALS; (ii) develop a high-throughput phenotypic drug screening assay to discover new compounds able to rescue a characteristic ALS phenotype: TDP43 protein aggregation. The discovery of new ALS treatments has been partly hampered by the complexity and heterogeneity of the disease. The first result chapter addresses these issues using an in-silico approach. Specifically, Weighted Gene Co-expression Network Analysis (WGCNA) was used to analyse a published transcriptomic dataset of patient post-mortem brain samples to identify causal relationships between co-expression networks and disease phenotypes and ultimately identify new druggable targets. Using this framework, it was found that ALS patients with bulbar onset presented a strong activation of oligodendroglial gene co-expression networks, which was not observed in the limb onset cohort. Further orthogonal validation in post-mortem tissue from a different ALS cohort revealed that sporadic ALS patients with bulbar onset had a significantly higher number of differentiated oligodendrocytes in the motor cortex. Overall, these observations implicate a particular role for oligodendrocyte dysfunction in bulbar onset ALS. Cytoplasmic accumulation of misfolded and aggregated TDP43 (transactive response DNA-binding protein 43kDa) in the brain and spinal cord of ALS patients is the pathological hallmark of the disease and found in the vast majority of the ALS population, sporadic and familial. In this work, CRISPR/Cas9 technology was used to generate a HEK293 cellular model carrying a pathogenic mutation on the gene that encodes TDP43 which, under oxidative stress, developed cytoplasmic TDP43 aggregates. This work served as the basis for the development of a phenotypic drug screening assay to identify small molecules able that ameliorate TDP43 protein aggregation. The screening of a library of 6652 small molecules (including FDA-approved drugs, investigational compounds, and natural products) led to the identification of several promising compounds that were taken forward for further validation. Following the primary drug screening, hit compounds were tested across a suite of assays designed to verify their activity in stem cell-derived spinal motor neurons and to confirm lack of neuronal toxicity. Finally, the last chapter explores the mechanism of action of the active compounds through a combination of bioinformatic tools and in-vitro experiments. Collectively, this thesis presents the development of a robust platform for the discovery of new disease targets and small-molecule therapeutics for ALS.

Finding and understanding patterns in gene expression guides our understanding of living organisms, their development, and diseases, but is a challenging and high-dimensional problem as there are many molecules involved. One way to learn about the structure of a gene regulatory network is by studying the interdependencies among its constituents in transcriptomic data sets. These interdependencies could be arbitrarily complex, but almost all current models of gene regulation contain pairwise interactions only, despite experimental evidence existing for higher-order regulation that cannot be decomposed into pairwise mechanisms. I set out to capture these higher-order dependencies in single-cell RNA-seq data using two different approaches. First, I fitted maximum entropy (or Ising) models to expression data by training restricted Boltzmann machines (RBMs). On simulated data, RBMs faithfully reproduced both pairwise and third-order interactions. I then trained RBMs on 37 genes from a scRNA-seq data set of 70k astrocytes from an embryonic mouse. While pairwise and third-order interactions were revealed, the estimates contained a strong omitted variable bias, and there was no statistically sound and tractable way to quantify the uncertainty in the estimates. As a result I next adopted a model-free approach. Estimating model-free interactions (MFIs) in single-cell gene expression data required a quasi-causal graph of conditional dependencies among the genes, which I inferred with an MCMC graph-optimisation algorithm on an initial estimate found by the Peter-Clark algorithm. As the estimates are model-free, MFIs can be interpreted either as mechanistic relationships between the genes, or as substructures in the cell population. On simulated data, MFIs revealed synergy and higher-order mechanisms in various logical and causal dynamics more accurately than any correlation- or information-based quantities. I then estimated MFIs among 1,000 genes, at up to seventh-order, in 20k neurons and 20k astrocytes from two different mouse brain scRNA-seq data sets: one developmental, and one adolescent. I found strong evidence for up to fifth-order interactions, and the MFIs mostly disambiguated direct from indirect regulation by preferentially coupling causally connected genes, whereas correlations persisted across causal chains. Validating the predicted interactions against the Pathway Commons database, gene ontology annotations, and semantic similarity, I found that pairwise MFIs contained different but a similar amount of mechanistic information relative to networks based on correlation. Furthermore, third-order interactions provided evidence of combinatorial regulation by transcription factors and immediate early genes. I then switched focus from mechanism to population structure. Each significant MFI can be assigned a set of single cells that most influence its value. Hierarchical clustering of the MFIs by cell assignment revealed substructures in the cell population corresponding to diverse cell states. This offered a new, purely data-driven view on cell states because the inferred states are not required to localise in gene expression space. Across the four data sets, I found 69 significant and biologically interpretable cell states, where only 9 could be obtained by standard approaches. I identified immature neurons among developing astrocytes and radial glial cells, D1 and D2 medium spiny neurons, D1 MSN subtypes, and cell-cycle related states present across four data sets. I further found evidence for states defined by genes associated to neuropeptide signalling, neuronal activity, myelin metabolism, and genomic imprinting. MFIs thus provide a new, statistically sound method to detect substructure in single-cell gene expression data, identifying cell types, subtypes, or states that can be delocalised in gene expression space and whose hierarchical structure provides a new view on the semantics of cell state. The estimation of the quasi-causal graph, the MFIs, and inference of the associated states is implemented as a publicly available Nextflow pipeline called Stator.

Immune checkpoint blockade has provided a revolution in solid cancer treatment particularly in the metastatic setting. Although many patients demonstrate clinical responses upon receiving checkpoint therapy, a considerable number of individuals fail to benefit from treatment and continue to progress from their cancers. It is therefore important to establish the immunological contributors that mediate treatment success and identify easily accessible biomarkers that can be used to monitor this process. Here, I utilised bulk and single-cell RNA sequencing, and flow cytometry, to characterise early peripheral immunity across immune checkpoint blockade in an Oxford-based cohort of metastatic melanoma patients. Firstly, I focus on peripheral CD8+ T cells as central mediators of anti-tumour responses. I demonstrate that a central tenet of the immune response to immune checkpoint blockade is mitosis, which is accompanied by type I interferon and interferon-G signalling and other immune activation pathways. I then employ single-cell RNA sequencing to dissect the CD8+ T cell compartment, finding that mitotically active cells are under-represented amongst clonal expansion. Instead, cytotoxic effector cells and expanding effector memory cells dominate the clonal pool and demonstrate phenotypic and clonal stability across immune checkpoint treatment. Immune checkpoint blockade shifts T cell subsets towards increased differentiation and has particularly stimulatory effects on effector CD8+ T cells. Next, I describe how these aspects of peripheral CD8+ immunity contribute to clinical patient responses and survival. Individuals with greater levels of CD8 clonal expansion and effector responses show improved treatment outcomes, whilst mitosis and interferon activity have limited patient benefit. I describe how a simple, tractable measure of clinical lymphocyte count predicts long-term survival benefit for patients. Lastly, I dissect how CD8+ T cell responses interact with other immune subsets, such as monocytes, B cells and CD4+ T cells to create expression and subset composition networks that reflect enhanced T cell activity and homeostasis, or dysregulated inflammation and suppressed T cell immunity. Intersecting these responses with genetic variation provides further insight into the factors influencing immunotherapy responses. Altogether, these results generate multiple blood-based biomarkers to predict therapy success, and provide novel angles for therapeutic adjuvants to support immune checkpoint blockade, spanning potential cellular, cytokine and metabolic mechanisms.

Infectious diseases have been a major cause of mortality and morbidity throughout human history. The dramatic advances in technology, especially in genome sequencing, have opened a new avenue to study infectious diseases and changed the landscape of public health and medical science. Compared to non-communicable human diseases where human genetics is the sole focus of genetic studies, infectious diseases add another layer of complexity due to the host-pathogen dynamic. Since infection can be seen as the result of the complex interplay of host and pathogen and the environment, a comprehensive understanding requires the incorporation of all these factors. In this thesis, I explore the different applications of genetic data in studying infectious diseases through the lens of host-pathogen dynamics and their clinical or public health implication. I first used virus whole genome data to rebuild the epidemiological history of Hepatitis C virus subtype 3a (HCV-3a). I showed how World War II had likely led to the spread of HCV-3a from South Asia and becoming a global epidemic. Notably, I also incorporated host genetic data into the analysis and showed how it helped us identify and address the confounding factor of human migration when conducting an evolutionary analysis on HCV-3a. I then integrated previously published methods to investigate protein residue covariation to investigate the signals co-evolution within the Hepatitis B virus (HBV). I built a co-evolution profile for HBV and link them to previous studies to demonstrate how drug resistance is one of the major driving forces of protein residue co-evolution in HBV. Lastly, I explored the potential of integration of national-scale databases that collect pathogen (Second Generation Surveillance System) and host data (UK Biobank) respectively and conducted multiple genome-wide association studies (GWASs) and lay out how such integration can help improve our understanding of human genetic risk factors for infectious diseases. To conclude, I discuss the importance of host-pathogen data integration as well as the need for developing statistical models that can account for both the data volume as well as complexity resulted from such integration.

Cyanobacteria represent a treasure trove of uncovered natural products (NPs) and unbeknown biosynthetic machinery. Currently the exponential rise in genome sequencing of cyanobacteria and other organisms has revealed a wealth of biosynthetic gene clusters, many of which cannot be linked to a NP. This work describes the investigation of cyanobacterial strains for the presence of non-ribosomal peptide synthase (NRPS) and polyketide synthase (PKS) genes screened by PCR and Sanger sequencing. In addition to whole genome sequencing (WGS) using Illumina and Oxford Nanopore was undertaken of selected strains. Bioinformatic tools were used to detect biosynthetic gene clusters (BGCs) and identify regions of interest. Furthermore, publicly available data, in the form of publications and nucleotide data from genome assemblies was gathered to form several datasets to analyse and link cyanobacterial metabolites to their genomes, and to uncover the diversity of cyanobacterial NPs. Recent development has been undergoing to establish heterologous expression tools for cyanobacteria. Currently only a dozen cyanobacterial natural products have been heterologously expressed, this work details the heterologous expression of a ribosomally synthesised and post-translationally modified peptide (RiPP) named viridisamide A from Oscillatoria nigro-viridis PCC 7112 using the DiPaC method. This thesis also describes the use of this method for the cloning of the Fischerazole BGC from Fischerella sp. 9431. Furthermore, the NP cyclomarin A, a marine natural product from an actinomycetes rather than a cyanobacterium, which possesses potent bioactivities against both tuberculosis and malaria was investigated. Here precursor directed biosynthesis of the fermented strain Streptomyces sp. BCC41611 was used to created halogenated cyclomarin variants. In addition, synthetic chemistry methods were used to functionalise the epoxide of the biosynthesised NP by azidolysis and copper(I) catalysed alkyne-azide cycloaddition. Lastly, the halogenase VirX1 from the cyanophage syn10 was studied. This phage infects the marine cyanobacterial genera Synechococcus and Prochlorococcus which are responsible for over a quarter of global photosynthesis. Here, the halogenase was investigated in order to attempt to uncover its natural substrates.

Pulmonary arterial hypertension (PAH) is a rare and fatal lung disease. To date, in only a third of idiopathic patients, the cause can be attributed to rare genetic variation in the protein-coding space. The sequencing of 13,343 whole genomes by the NIHR BioResource for Translational Research - Rare Diseases (NBR), including 1,216 PAH samples, provides an unprecedented opportunity to estimate the contribution of regulatory genome variation to the development of PAH. This work aims to determine whether sequence-based predictions of epigenetic features can be used to narrow down the possible regions of interest and allow aggregation of variants into functional groups for association testing. A convolutional neural network (CNN) has been trained using publicly available data sets to predict epigenetic features from DNA sequences. The model was tested against known enhancer regions and its accurate performance was verified; two approaches were developed for the evaluation of the epigenetic features. Firstly, an epigenetic importance score supplies general information about the availability of epigenetic profiles within a region to explore the non-coding space. Secondly, a regulation score combines the predicted features into activating and repressing subsets for more detailed analyses to gauge the regulatory impact of variants. Based on the regulatory impact and other common variant annotations, variants were filtered and aggregated for over-representation analysis, comparing cases with controls. These scores were used in an outlier analysis using the Fisher's exact test and a sweep across the landscape of 1,135 PAH gene-associated enhancers, using SKAT-O. After p-value adjustment, over 80 regions were found significant. The statistical analysis revealed likely disease-causing sequence variation in ENG enhancers, as well as strong associations in ACVRL1 and KLK1 enhancers. Here, I present an extended search into enhancer networks associated with PAH, unlocking the non-coding space for genomic medicine.

Bacterial phylogenies often contain distinct clusters or "clades", associated with different ecological niches despite frequent horizontal gene transfer (HGT) between clades. An excellent example is the Bacillus cereus (Bc) group, whose members share common adaptations but are often found in different clades, raising questions over each clade's specific ecology and pathogenic potential. Identifying consistent genetic and phenotypic differences between clades within the Bc group will determine whether clades truly possess specific ecologies and reduce confusion regarding the pathogenic potential of the different clades. The ecology of each Bc clade was explored through identification and comparison of genes under selection within the core and accessory genomes within the three largest clades within the Bc group. Each clade had its own distinctive signature of selection, with Clade 3 showing signatures of adaptation to low temperatures, supporting the hypotheses of previous studies. Experimental evolution explored the effect of different thermal niches on the ability of a Clade 2 B. thuringiensis and Clade 3 B. mycoides strain to adapt to novel environmental conditions. After 700 generations of adaptation to a novel growth medium, fitness gains in each strain were found to be highly dependent on temperature. Single nucleotide polymorphism (SNP) variant analysis of evolved strains indicated each strain responded differently to each temperature, suggesting adaptation in these strains was constrained by thermal niche. Inactivation of a putative thermal niche defining gene, cspA, in B. thuringiensis caused the fitness of the strain to increase at low temperatures contrary to predictions, with different levels of temperature dependence in variants of the gene from mesophilic and psychrotolerant strains. However, neither inactivation of the cspA gene or complementation with the psychrotolerant variant affected the B. thuringiensis strain's ability to adapt during a second experimental evolution experiment. The phenotypic inertia of thermal niche within members of the Bc clades suggests that thermal niche may form a basis for the maintenance of distinct clades within the Bc group. The reason for thermal niche constraint is unclear, but further study can provide insight into how clades with distinct ecologies are maintained over time.

Neural stem cells (NSCs) in the brain can be reactivated to exit quiescence and generate new cells in response to injury or disease. Determining the specific signals involved in NSC reactivation is essential to understanding neural regeneration. Investigating NSCs in *Drosophila melanogaster* provides a powerful model due to ease of identification of quiescent NSCs and a plethora of genetic and molecular tools available to study the nervous system *in vivo*. During development, *Drosophila* NSCs reactivate in response to dietary amino acids sensed by the fat body, a sensor organ analogous to liver and adipose tissue in mammals. Previously unknown signals from the fat body are transmitted to the bloodbrain barrier (BBB) glia on the surface of the brain. The glia secrete insulin-like peptides, which are received by the insulin receptor on NSCs, inducing their reactivation. To identify the fat body-derived signals sent to the brain, I used Targeted DamID to generate transcriptional profiles of the fat body under fed and starved conditions. I discovered upregulation of *dpp (decapentaplegic)*, a transforming growth factor (TGFβ) secreted morphogen, after feeding. Knockdown of Dpp in the fat body severely impaired NSC reactivation. Similarly, knocking down a key receptor for Dpp, Tkv (Thickveins), in the BBB glia led to severely impaired NSC reactivation. To compare gene expression in glial subtypes under fed and starved conditions, I generated single-cell RNA sequencing datasets. I found Dpp to be upregulated in BBB glia in response to feeding and functional experiments showed that Dpp signalling in the glial niche is required for NSC reactivation. The presence of Medea transcription factor binding sites at the dIlp6 locus and impaired NSC reactivation when Medea is knocked down in the BBB glia suggest that Dpp signalling may regulate expression of dIlp6. Knockdown of Tkv in the NSCs themselves did not affect reactivation, suggesting autocrine Dpp signalling in the BBB glia in response to interorgan Dpp signalling from the fat body to the brain. The findings may help understand how to induce neural stem cell proliferation after brain damage or neurodegeneration.

Adenosine triphosphate binding cassette (ABC) transporters form a ubiquitous superfamily of integral membrane proteins involved in the translocation of substrates across membranes. Substrate translocation is powered by ATP. In this study, four medically important classes of ABC transporters were chosen for structural studies: namely, subfamily B, member 5 (ABCB5), subfamily B, member 6 (ABCB6), subfamily G, member 1 (ABCG1), subfamily G, member 4 (ABCG4). A bioinformatics approach was used for ortholog selection of representative ABC transporters to improve the process of structure determination. Selected orthologs were expressed in Saccharomyces cerevisae with polyhistidine affinity and Green Fluorescent protein identification tags and purified in the presence of the detergent, n-Dodecyl-β-D-Maltoside. A battery of tests was applied to explore the quality of purified proteins. An assay using a thiol-specific dye, and intrinsic protein fluorescence was used to assess the thermal stability of proteins. ATPase activity, measured in a detergent environment, showed that all of the proteins hydrolysed ATP. A membrane thermal shift experiment demonstrated that mouse ABCG4 was stabilized by tyrosine kinase inhibitors. Mouse ABCG4 was also the subject of cryoEM imaging. Taken together, this project throws light on the use of computational tools for target selection before conducting structural studies. All target proteins were successfully expressed in yeast. Protein purification showed mixed results; however, all proteins could be enriched. All purified protein targets were functionally active. The assays could form a platform for screening compounds that stabilise the proteins. Stabilized proteins could subsequently go into crystal trials.

A genome-wide association study (GWAS) is a standard approach for understanding the relationship between genetic variants and disease, resulting in a better understanding of disease mechanisms. However, despite the vast contribution of GWAS to our knowledge and understanding of human biology, our ability to interpret these findings has not yet been fully developed. This is primarily because most disease-disposed regions are found within the non-coding regions of the genome, which affect regulatory elements such as enhancers and promoters. There is currently no clear path to identify causal regulatory SNPs and to link these to the genes and cell types they affect. This thesis work focuses on first understanding scATAC-seq as a technology and its outputs and then taking advantage of its potential power to prioritise non-coding genetic variants and cell types. I have deeply reviewed and explored single-cell open chromatin epigenetics, which led to discovering the effect of the Tn5 cutting pattern on providing different levels of information from scATAC-seq. I developed a mathematical strategy to remove nucleosomal background from scATAC-seq data, increasing the data resolution and enabling the accurate identification of TF-bound regions. We successfully built an analytical platform, Avocato, which provides a complete solution to understand the relationship between non-coding genetic variants and their affected diseases. Avocato is not only capable of analysing and visualising scATAC-seq data but also removes background noise resulting in high-resolution data to be used in prioritising non-coding genetic variants and cell types. Additionally, Avocato allows users to interact with their analysis and prioritisation results interactively in a user-friendly interface. Together this thesis builds on existing technologies, bringing us one step closer to fully understanding GWAS findings and functional studies and elucidating the mechanisms of common human diseases.

Gene and genome duplications have an important role in evolution as a mechanism that creates novel genetic material. Paralogues created by duplication can diversify in function and contribute to the complexity of genetic networks, especially in the context of the gene regulation and signalling pathways that control animal development. Clusters of related genes that have arisen by tandem duplication illustrate the role of regulatory elements in preserving synteny by constraining gene neighbourhoods. Whole genome duplications have occurred on the stems of lineages characterised by the evolution of novel structures, adaptations to different environments, and species diversity. Here, I aim to understand how gene and genome duplications have impacted specific developmental gene families and processes throughout animal evolution with a comparative genomics approach. I make use of genomic resources from phylogenetically informative lineages, including new genomes of early-branching chordates, spiralians, and ecdysozoans. I have examined the impact of the vertebrate two rounds of whole genome duplication on chordate muscle development including the highly conserved family of myogenic regulatory factors and muscle gene expression. I have also investigated the evolution of homeobox gene families, with a focus on clusters of genes found across bilaterians, and specifically focus on the Hox cluster. With phylogenetic and synteny analyses, I found that a tandem duplication underpins the origin of two vertebrate muscle gene types. I examined the impact of whole genome duplication on a diversity of genes encoding proteins in a highly conserved signalling pathway in muscle, and I surveyed and revised the hypotheses for the evolution of homeobox genes across the bilaterians. In doing so, I have also generated a large resource of genomic annotations and protein sequences to facilitate functional studies in the future. These findings not only have identified certain duplications that have underpinned certain known instances of subfunctionalisation among paralogues, but also indicate gene families to target for future study.

Tuberculosis killed an estimated 1.5 million people in 2020 and the spread of drug-resistant strains is an increasing threat. Fluoroquinolones are among the safest and most effective drugs used to treat drug-resistant tuberculosis, but fluoroquinolone resistance has emerged. Rapid resistance diagnosis is key to provide patients with effective treatment and monitor the spread of resistant strains. Sequence-based tools, including whole genome sequencing (WGS), provide a promising solution whereby known resistance associated mutations can be quickly detected and used to infer resistance. Several catalogues of mutations with known associations have now been produced, however, these catalogues will never be exhaustive owing to the inevitable emergence of novel and rare resistant mutations. In this thesis I explore a geographically diverse dataset of thousands of Mycobacterium tuberculosis isolates with WGS and phenotypic resistance measurements, to improve our understanding of patterns associated with fluoroquinolone resistance. I investigate the reliability of the assumptions of current sequence-based diagnostics and show how resistance patterns can affect the performance of predictive tools. I apply and evaluate two methods, that utilise the chemical and structural interactions resulting from genetic mutations, to predict the effects of mutations on fluoroquinolone resistance; machine learning algorithms and free energy calculation. I find that fluoroquinolone resistance is widespread and the associations with resistance are complex; genetic and geographic background, resistance to other antitubercular drugs and minor populations with resistance associated mutations are important. I show that structure-based predictive methods, and the machine learning approach particularly, can successfully predict fluoroquinolone resistance and susceptibility. Such tools could increase the success of resistance prediction from WGS data by complimenting the catalogue-based predictive approach and predicting the effects of novel mutations. Overall, this work shows that genetics-based predictive diagnostics have the potential to provide personalised, effective treatment regimens for tuberculosis.

Characterising microbial communities enables a better understanding of their complexity and the contribution to the environment. Metagenomics has been a rapidly expanding field since the revolution of next generation sequencing began, and it has a wide range of application including for medicine, agriculture, forensics, archaeology and even domestic use [Sarkar et al., 2021, Holman et al., 2017, Khodakova et al.,2014, Santiago-Rodriguez et al., 2017, Vilanova et al., 2015]. Sequencing amplicon data, such as 16S rRNA, is now commonly used to characterise the microbiome in a variety of biological samples. However, their correct taxonomic identification still remains a challenge, and often short reads are identified, correctly or not, at several ranks of the taxonomic tree other than species or subspecies level. Every metagenomic study is designed for specific needs, and it is often complicated to find a suitable bioinformatics pipeline and reference database. There is currently a lack of systematic benchmarking of in-house methods for metagenomics. The work presented in this thesis aims to establish an approach for the in silico validation of 16S rRNA metagenomic data. A method to generate realistic in silico metagenome data that resembles project-specific sequencing data is presented, including a new process to generate synthetic negative controls for amplicon data, which can be employed regularly to assess the appropriateness and optimisation of methods for specific metagenomic projects. To aid the benchmarking process, new metrics have been defined based on a measure of taxonomic distance. A k-mer based method with the lowest common ancestor approach was selected to investigate a range of factors that influence meta-taxonomic classification success. It includes the comparison of database quality filtered at various levels, and as well as a comparison of different taxonomic annotation methodologies. The experimental findings reveal the importance of having highly curated taxonomic annotations of the genetic sequences in the database, and that a missing fraction of the tree of life can lead to misclassification of any related or unrelated organisms. In some cases, it is shown that longer reads can help to improve assignment, with mutations and sequencing errors having a relatively low negative impact. The marker gene 16S rRNA has well-defined conserved and variable regions, which help to distinguish species. Therefore, these regions were studied and also recalculated using information theory, to investigate which parts of the sequence are discriminative for metagenomic taxonomic identification. In addition, linguistics methods, Term Frequency - Inverse Document Frequency (TF-IDF) coupled with multinomial naive Bayes, is shown to provide understanding of genetic signatures and is applied to generate a new method to classify taxonomically metagenomics short reads. Biological samples were taken from cattle respiratory tract, DNA was extracted and sequenced to provide metagenomic data. Two sets of experiments were carried out, (i) to compare sampling and extraction methods and (ii) to characterise the microbial community observed in young cattle in the different lung lobes and nose. The data reveal that the composition of the microbial community observed is highly dependent on the sampling method.

Cyanobacteria are an ancient clade of photosynthetic prokaryotes, varying in morphology, physiology, biochemistry and habitat. They evolve by typical prokaryotic mechanisms including horizontal gene transfer (HGT). Some species produce toxins (cyanotoxins) that present health hazards to humans and animals, with potential harm to local economies. The biosynthetic pathways and roles of some cyanotoxins are unclear. The rapid increase in high quality publicly available genomes presents opportunities for discovery from comparative genomics in cyanobacteria. The work presented here focuses on three topics in cyanobacteria, using bioinformatics analyses of 130 cyanobacterial genomes. Firstly, I consider hypotheses for the biosynthesis and physiology of the non-encoded neurotoxin 2,4-diaminobutanoic acid (2,4-DAB). Secondly, I consider hypotheses for the biosynthesis and potential roles of its structural analogue, β-N-methylaminoalanine (BMAA). These topics use similar methodology: pairwise and multiple sequence alignment, profile hidden Markov models, substrate specificity and active site identification, and the reconstruction of gene phylogenies. We show that some species have genes involved in known biochemical pathways to 2,4-DAB - genes coding for proteins in the aspartate 4-phosphate pathway (including the diaminobutanoate-2-oxo-glutarate transaminase, the downstream decarboxylase, diaminobutanoate decarboxylase, and ectoine synthase) - and BMAA (homologs of the Staphylococcus aureus genes sbnA and sbnB). We highlight the possible involvement of 2,4-DAB and BMAA in the production of siderophores. We show that the biosynthesis of 2,4-DAB and BMAA is likely to be confined to a limited number of species, or to occur via different, unknown, pathways. Thirdly, I investigate hypotheses concerning the association of HGT events with environmental context. I test existing hypotheses claiming that genetic exchanges are more frequent in extreme habitats (versus mesophilic) and in terrestrial habitats (versus aquatic). My results, based on reconciliation of gene trees with the species tree, do not suggest a link between the prevalence of HGT and extreme or terrestrial environments. I highlight the need for complete descriptions of the isolation source and culture type (axenic, non-axenic monocyanobacterial culture, environmental sample), the need for accurate and robust methods for HGT inference, and for more objective and detailed criteria for environmental classification and of cyanobacterial species. This work contributes to research into cyanobacterial neurotoxins and provides insights into the prevalence and distribution of HGT in cyanobacteria.

Single-cell sequencing technologies have unsurprisingly become a favourable choice for studying key biological questions about cell heterogeneity, rare cell types or lineages. It is only cell-level resolution that allows for an accurate analysis of internal cell processes such as mutagenesis. Eventually, single-cell RNAseq could provide an explanation of mechanisms that lead to the ultimate transformation of healthy tissues into cancerous lesions. One of the main interests of my lab is Barrett's oesophagus. It is a highly clonal disease and a likely cancer precursor. We decided to take advantage of the single-cell RNAseq technology in order to attempt to identify the tissue of origin of the disease which, despite years of research, still remains unknown. However, the range of methods for identification of mutations in single cells is very limited. In order to address that, we developed our own single-cell RNAseq variant caller. We validated it on a publicly available breast cancer dataset by achieving a reasonable intersection of our results with the output of commonly used bulk tools. Furthermore, we showed that our caller was capable of identifying expected data characteristics such as known breast cancer signatures and mutations in breast cancer genes. We then applied our method to the Barrett's dataset to investigate connections of Barrett's with surrounding tissues. Contrary to the previous transcriptomic analysis conducted on the same dataset and indicating a Barrett's-oesophagus connection, our results revealed a more likely link of Barrett's with the stomach.

In 1933 the German geneticist Richard Goldschmidt presented his theory that major evolutionary events have been driven, not by the slow accumulation of small variants, in line with Darwinian thought, but instead large-scale mutations which he called macromutations. His theory was called 'The hopeful monsters hypothesis', as any species arising from such an event would often result in evolutionary dead end. This theory was, however, by and large rejected due to its stark contrast with that of Darwin's theory of evolution. 86 years on, thanks to tremendous advancements in genomic technologies and evolutionary analysis, we are aware of the existence of a myriad of what could be considered macromutations (large-scale duplications) in the human genome, and the significant role they have played in our evolution. Whilst beneficial in the process of shaping who we are as a species, these large-scale mutations are not always benevolent, having been found to have significant links with a plethora of human diseases. Using evolutionary data, alongside important biological features of human genes, we explore the links between large-scale mutations, particularly whole genome duplications, introgressed genes, and heritable, parasitic, and viral disease. We find that there are strong links between ancient fragility, exposed by divergence between duplicated genes, and heritable disease; that co-evolution between humans and the soil borne helminth parasite Trichuris trichura has been facilitated by strong selective pressures exerted on the ancient genes of the TGFb superfamily; and that viruses target ancient and important gene families and functions. It is clear that large-scale mutations, in particular gene and genome duplications, have provided the genetic redundancy that contributed to the evolution of the complex species we are today. However, given the fragility and propensity of genes of this kind to be associated with disease of all varieties, our future evolvability, and ability to adapt is in question. We must, therefore, readdress Goldschmidt's theory as a question - Are we merely the descendants of hopeful monsters?

Background: The disease course of rheumatoid arthritis (RA) varies widely between patients, and various therapeutic options are available. However, none are universally effective, all have a risk of side-effects and currently, all are prescribed on a "trial and error" basis, based on escalating cost and not precision medicine targeted to patient endotype. Previous work has shown that circulating drug concentration levels of tumour necrosis factor inhibitors (TNFi, a class of drug used to treat RA and other autoimmune conditions) are associated with response to treatment. This thesis hypothesised that biological factors, such as protein expression, contribute to variability in circulating drug levels and treatment response to biologic agents in patients with RA. Methods: A population pharmacokinetic (popPK) study was carried out in patients with RA starting either Amgevita or Benepali, which are biosimilar agents for the TNFi agents adalimumab and etanercept, respectively. Model parameter estimates from the popPK study were used to simulate altered dosing intervals of these drugs. Proteomics data was obtained on all patients in the popPK study, as well as an additional cohort of patients with RA starting etanercept, using Sequential Window Acquisition of All Theoretical Mass Spectra (SWATH-MS). Protein expression was regressed against RA clinical outcome measures to determine any associations between protein expression and treatment response. Protein expression was also analysed alongside paired genotype data to determine whether any protein quantitative trait loci (pQTLs) existed. Significant pQTLs were then used to construct a polygenic risk score (PRS) for treatment non-response. Results: 16 patients were recruited to the popPK study; PK parameters were successfully estimated and used to simulate the effect of altered dosing intervals. SWATH-MS was used to generate proteomics data in serum samples from 180 selected patients commencing etanercept recruited to the Biologics in RA Genetics and Genomics Study Syndicate, a prospective multi-centre UK-based observational cohort. Proteomics analysis identified 52 proteins associated with RA clinical outcome measures. A pQTL analysis was carried out using 147 patients from the etanercept sub-cohort. 104 pQTLs were identified, 14 of which overlapped with significant proteins from the regression analysis. A PRS was generated using significant pQTLs, but was not found to be statistically significantly predictive of poor treatment response. Conclusions: The popPK study has provided proof-of-concept for future personalised dosing trials in patients with RA starting TNFi. This thesis has identified several proteins associated with RA clinical outcome measures that also have a genetic basis. Findings require external validation with replication studies in an independent cohort, but once confirmed, this could pave the way for future biomarker and/or drug target development.

Each human exome contains thousands of protein-altering variants located in more than 19,000 genes. Humans typically have two copies of a gene, and variants that affect one or both gene copies are called heterozygous and homozygous, respectively. If one gene copy is affected by deleterious heterozygous variation and cannot produce normal protein, this could result in a dominant disease. However, some genes can tolerate disruption of one copy, but deleterious homozygous or two heterozygous variants in different copies could still result in a recessive disease. Finally, humans can tolerate the inactivation or deletion of both copies of some genes without developing diseases. Because studied diseases' inheritance patterns are frequently known (e.g. if one of the parents and a child both have a disease, the inheritance pattern is likely to be dominant), clinical researchers want to know a candidate disease-causing variant inheritance pattern to prioritise candidate disease genes for laboratory validation. Although inheritance pattern is a property of disease causing variants, it can be predicted using gene-level properties. The aim of this study was to develop gene-level computational metrics that can be used for this task, and recently created large variant population databases such as Genome Aggregation Database (gnomAD, >137,000 individual exomes/genomes) provided novel data for such studies. This thesis is written in the journal format and consists of three paper-style result chapters. In the first paper, we analysed deviations from Hardy-Weinberg Equilibrium of rare variants in gnomAD to detect potential disease-causing and heterozygous advantageous variants based on homozygous deficiency in the healthy populations. The second paper developed a gene variation intolerance ranking (GeVIR) system by measuring how unevenly variants in gnomAD were distributed in a gene relative to other genes. Finally, in the third paper, we developed multiple supervised machine learning models based on various gene properties (including GeVIR) and combined them into a single continuous gene ranking metric that can be used to measure gene predisposition to disease inheritance patterns (DIP). In conclusion, this thesis contributed to the understanding of variant population data and the application of supervised ML methods to classify candidate disease genes in the context of disease inheritance patterns. The primary outcome of this research was the development of two continuous gene metrics, GeVIR and DIP (available for 19,361 and 15,794 protein-coding genes, respectively), both of which can be used to distinguish dominant, recessive and non-disease genes. We anticipate that these metrics will aid clinical researchers in the prioritisation of candidate disease genes.

Intra-cellular signalling determines how cells process information. Through the integration of diverse chemical and physical stimuli, cells can enact transcription, among other changes to modulate growth, fate and survival. It is through the dysregulation of such processes that many diseases, including cancer originate. For many years, our study of signalling processes has been based on discrete 'pathways', characterised mostly by small-scale studies. However, as more system-wide data becomes available, it is becoming increasingly obvious that intra-cellular signalling is more like a dense and inter-connected network, with intense and functional cross-talk between pathways. In my PhD project, I utilised both new 'omics' data and a plethora of network-based approaches to guide our understanding of various signalling-related disease contexts. Networks provide a convenient framework with which to explicitly control the level of prior-knowledge required to understand complex 'omics' data. Escaping the study bias that has so-far dominated our characterisation of intra-cellular signalling is vital if we are to progress in our understanding of complex cellular behaviours.

Background: Melioidosis is caused by the Gram-negative bacterium Burkholderia pseudomallei, and is a leading cause of death from community-acquired infections in Northeast Thailand with a case fatality rate of up to 50%. Most melioidosis patients have one or more comorbidities, and more than half of patients have diabetes which confers a 12-fold increased susceptibility to melioidosis. A similar, but less marked, relationship is seen in tuberculosis (TB) patients, with a 3-fold increased risk of developing TB in people with diabetes. Little is known regarding the mechanisms underlying the increased susceptibility in diabetes to these intracellular pathogens. Likewise, the host response associated with fatal melioidosis remains to be fully elucidated. Methods: Whole blood RNA sequencing data were generated from three independent melioidosis cohorts from Northeast Thailand and one global multi-site tuberculosis study. Both supervised and unsupervised data analysis approaches were performed including differential gene expression (DGE) analysis, pathway analyses, upstream regulator analysis, weighted gene co-expression network analysis (WGCNA), and hub gene analysis. Results: Firstly, principal component analysis (PCA) revealed clear separation between fatal and surviving melioidosis patients. DGE analysis identified profound changes in the transcriptome of fatal melioidosis cases with over 600 up- and 1,000 down-regulated genes compared to survivors. Pathway analyses following DGE analysis identified exaggerated inflammation and up-regulated type 2 immune responses concurrent with massively down-regulated T cell signalling cascades in fatal melioidosis. Upstream regulator analysis identified potential key regulators associated with fatality during melioidosis including TNF, IL4, IL1A, PTGER2, and OSM. Likewise, the unsupervised analysis by WGCNA confirmed that several co-expressed gene modules enriched for inflammatory immune responses driven by the innate immune compartment were associated with 28-day mortality. Furthermore, hub genes associated with fatal melioidosis such as CEACAM1, SERPINB1 and SORT1 may play deleterious roles during melioidosis. Secondly, the impact of diabetes on the transcriptome was subtle in acute melioidosis, with similar but more pronounced changes in TB, which is a more chronic disease. Diabetes status was associated with enhanced non-specific inflammatory responses for both melioidosis and TB, alongside failure to fully switch on interferon-mediated responses for TB. Conclusion: The transcriptomic signature in fatal melioidosis is characterised by concomitant excessive inflammation, a pronounced anti-inflammatory response, and a stunted T cell response. These results highlight the pivotal role of adaptive immunity during fatal melioidosis. Furthermore, diabetes is associated with increased activation of innate pathways and reduced interferon responses which may contribute to the increased susceptibility observed for these intracellular pathogens.

In this work, statistical learning approaches are used to detect biomarkers for neurodegenerative diseases (NDs). NDs are becoming increasingly prevalent as populations age, making understanding of disease and identification of biomarkers progressively important for facilitating early diagnosis and the screening of individuals for clinical trials. Advancements in gene expression profiling has enabled the exploration of disease biomarkers at an unprecedented scale. The work presented here demonstrates the value of gene expression data in understanding the underlying processes and detection of biomarkers of NDs. The value of novel approaches to previously collected -omics data is shown and it is demonstrated that new therapeutic targets can be identified. Additionally, the importance of meta-analysis to improve power of multiple small studies is demonstrated. The value of blood transcriptomics data is shown in applications to researching NDs to understand underlying processes using network analysis and a novel hub detection method. Finally, after demonstrating the value of blood gene expression data for investigating NDs, a combination of feature selection and classification algorithms were used to identify novel accurate biomarker signatures for the diagnosis and prognosis of Parkinson's disease (PD) and Alzheimer's disease (AD). Additionally, the use of feature pools based on previous knowledge of disease and the viability of neural networks in dimensionality reduction and biomarker detection is demonstrated and discussed. In summary, gene expression data is shown to be valuable for the investigation of ND and novel gene biomarker signatures for the diagnosis and prognosis of PD and AD.

Plasmodium knowlesi is a zoonotic malaria parasite of Southeastern macaque monkeys that causes zoonotic malaria in humans. P. knowlesi is also an experimental model for malaria, and information on P. knowlesi largely stem from experimental lines first isolated >four decades ago, rather than contemporary isolates causing human infections. The experimental lines are laboratory-restricted and have become relatively genetically stagnant and free from the selection pressure that would naturally occur in nature. Within the P. knowlesi genome exist the Schizont Infected Cell Agglutination variant antigen (SICAvar) and Plasmodium knowlesi interspersed repeat (kir) multigene families significant, which are of biological and scientific interest. To provide context using contemporary clinical isolates, this project aimed to generate high-quality genome sequences using long-sequencing from clinical 'wild-type' samples from infected patients. This includes generating new information on variant multigene families in P. knowlesi genomes generated from clinical patient whole blood. The work presented here details a method to deplete leucocytes in thawed P. knowlesi-infected patient whole blood samples to generate parasite-enriched DNA for whole-genome sequencing, resulting in >95% human DNA reduction. The extracted DNA was sequenced with long-read sequencing technology to create de novo whole-genome assemblies. From these, two isolate genomes representing the two dimorphic clusters of P. knowlesi in clinical samples were analysed. The generated genomes are highly syntenic to the published reference genome, sharing >4500 orthologous clusters with the PKNH reference genome. However, the number of SICAvar and kir genes present in the dataset deviated from the published reference genomes of P. knowlesi. The successful generation and construction of these patient genomes aid further interrogation of the contemporary P. knowlesi genome, with a focus on the constituent genes present in comparison to the experimental line.

Rhizobium leguminosarum is a soil bacteria which infects plants of the legume family such as peas, lentils, and beans. When associated with the plant, R. leguminosarum fixes atmospheric nitrogen (biological nitrogen fixation), producing ammonium that is assimilated by the plant, improving plant growth. This symbiotic relationship is very important for agriculture; increasing the knowledge of plant infection and nitrogen fixation processes might lead to a decrease in the use of chemical fertilisers. In this thesis, I combine R. leguminosarum gene expression data from previous experiments to produce gene coexpression networks and increase the functional annotation of this bacterium. Gene coexpression networks are networks in which nodes represent genes and edges represent coexpression (i.e. two genes are connected if they have a similar expression pattern). Unfortunately, there is no standard method to generate gene coexpression networks from gene expression data. I introduce signed distance correlation as a measure of dependency between two variables and use it to generate self-consistent, unweighted and weighted, gene coexpression networks that are more stable and capture more biological information than those obtained from Pearson correlation or mutual information. Subsequently, I present a pipeline that combines novel scores from gene coexpression network analysis in a principled way to identify the genes that are associated with certain growth conditions or highly coexpressed with a predefined set of genes of interest. This association can lead to putative functional annotation or to a prioritised list of genes for further study. Lastly, I provide a detailed analysis of the transcriptome of R. leguminosarum. For this purpose, I generate different knock-out mutants lacking transcriptional regulators which are reported to be important in the nitrogen-fixing state of the bacteria and perform RNA-Seq experiments. I also carry out a preliminary study of the molecular signals that may affect the changes in gene expression between the free-living and the plant associated bacteria.

Recent advances in live imaging technology, such as the MS2-GFP system, have enabled the recording of transcriptional data at the single-cell level at ever-greater temporal and spatial resolution. Whereas previously researchers had to rely on static 'snapshots' of developing embryos, such as those provided by Single Molecule Fluorescence in situ Hybridisation (smFISH), it is now possible to record fluorescence microscopy movies of developing embryos in the laboratory. An example of one such live imaging technique is the MS2-GFP system, where gene editing is used to insert a transgene into a gene of interest. When the gene is transcribed, a noisy fluorescent time series is generated which acts as a proxy for transcriptional activity. The dorsal-ventral patterning system in the early Drosophila embryo provides an ideal system for studying transcription using live imaging. In this system, a single input, a member of the Bone Morphogenetic Protein (BMP) family, controls multiple target genes, each of which exhibit transcriptional bursting, where transcripts are produced stochastically in discrete 'bursts' of activity, rather than as a constant, Poissonian process. The aim of analysing these movies is to gain insight into transcriptional regulation in the early embryo, i.e. the relation between the dynamics of mRNA production and cell developmental fate. BMP signalling is of particular interest due to the known involvement of misregulated BMP signalling in developmental defects and cancer. A key problem is how to process and analyse MS2 datasets in order to answer this question. The main output of the thesis is the development of a novel type of Hidden Markov Model (HMM) for extracting kinetic parameters from MS2 movies, with the aim of establishing the relationship between BMP signalling and transcriptional bursting. We first provide an overview of BMP Signalling in Drosophila, followed by a summary of previous theoretical definitions of biological noise and transcriptional bursting in the literature. We then outline the details of the implementation of our algorithm. The algorithm demonstrates a significant improvement in computational efficiency relative to the current state of the art model for MS2 analysis, the Compound State Hidden Markov Model (cpHMM), while allowing for the inference of single-cell transcriptional parameters. Results are shown comparing our algorithm to the original algorithm in terms of computational speed and accuracy, using synthetic and experimental Drosophila data. Finally, we present the in-depth results from using our algorithm to investigate the bursting dynamics of the Drosophila ush and hnt genes. We have been able to establish that regulation of bursting dynamics in this system is achieved through frequency modulation, i.e. by regulating the frequency of bursts, rather than burst duration or amplitude; burst frequency decreases as a function of distance from the embryo midline.

Microorganisms are found in virtually all environments, and while the majority of microorganisms are often prokaryotes, by biomass there are suspected to be 6 times more prokaryotes than fungi globally (Bar-On et al., 2018), eukaryotes are also important constituents of microbial communities (e.g on the human skin). Shotgun metagenomics can provide access to the combined genetic information of a community in a culture independent manner. Using de novo assembly and post processing methods, it can lead to the generation of so called metagenomic assembled genomes (MAGs), which provide contextualised access to genes of these elusive organisms. As most studies have focused on the recovery of prokaryotic MAGs, I first examined the limitations and gaps of existing tools with respect to their ability to recover microbial eukaryotic genomes. This led to the development of EukCC, a software to estimate the completeness and contamination of eukaryotic MAGs. Evaluation of this software showed that it is well suited for the fully automated recovery of eukaryotic MAGs. This workflow was applied to dataset obtained from several biomes to recover eukaryotic MAGs. However, I also demonstrate that eukaryotic MAGs can sometimes be fragmented and developed a merging algorithm to create merged MAGs (mMAGs). With the implementation of this algorithm in EukCC 2, I search a large number of datasets from MGnify for known and novel eukaryotic MAGs. Completing the eukaryotic MAG recovery process, I discuss how species-level dereplication for eukaryotes can be approached based on the genetic information alone. In summary I show that recovery of eukaryotic MAGs is a challenging but can be largely automated allowing large-scale studies to be performed.

The understanding of a therapeutic compound's mechanism of action (MoA) can provide great benefits to the drug discovery process. Biological and chemical information e.g., transcriptomics data is increasingly becoming easier to measure by high throughput methods, and large amounts of this type of data are available in the public domain. Such information can be used with methods such as machine learning, causal reasoning and pathway enrichment to gain insights on compound MoA across different levels of biology. This presents an opportunity to investigate approaches for integrating information sources and computational methods for the elucidation of a systems-level view of compound MoA, which is the topic of this thesis. In the first chapter, causal reasoning algorithms were benchmarked for their ability to retrieve compound targets and mechanistic signalling pathways from prior knowledge networks. We found that these algorithms generally perform poorly for recovering targets; but are able to recover mechanistic pathways better than traditional pathway enrichment using gene expression data. We also found that algorithms perform differently with different networks, and that the connectivity of mechanistic proteins on the network or their biological function can greatly impact the ability of the algorithms to retrieve them. In the second chapter, we integrated biological and chemical information to generate MoA hypotheses for the tau aggregation inhibitor Anle138b. By using both chemical structure and transcriptomics information, along with causal reasoning, pathway enrichment and machine learning-based target prediction, we were able to generate hypotheses across multiple levels of biology. We found biologically plausible hypotheses which were highlighted across all three methodologies, e.g., cholesterol metabolism modulation, which is a known mediator of tau aggregation. In the final chapter, we developed an R/Shiny app to allow users to carry out integrated chemical structure-based target prediction with gene expression-based causal reasoning and visualisation in a GUI environment. We also provide a case study demonstrating its capabilities.

Analysing genetic variation in pathogen genomes is key to understanding their biology, evolution and epidemiology. Typically, this is done by assembling one arbitrary genome, defined as the 'reference', and describing other samples as deviations from it. However, this model breaks down in highly diverse regions of the genome, where sample sequencing reads, differing too substantially from the reference, fail to map. This 'bias against diversity', due to using a single reference, naturally affects genomic regions under pressure to diversify: this includes the human MHC, a motivating example for the field, and vaccine candidate genes in the malaria parasite Plasmodium falciparum (Pf ), the motivating example for this thesis. The growing solution in the field is to build graph-based models, representing not one, but a population of genomes from a species, and using these genome graphs as a substrate for read mapping and genotyping instead. In this thesis, I develop new algorithms and data structures for genotyping highly variable genes in Pf, using genome graphs. In Chapters 2 and 3, I describe methods and code to analyse variation in highly diverse regions of the genome, across many genomes in a cohort. In doing so I provide two main advances on the state-of-the-art: jointly studying small (SNPs, indels) and large (indels >50bp) variation, and accessing variation on multiple references. I validate these methods using different datasets, ultimately genotyping SNPs on diverged haplotypes in two highly variable Pf genes, including one gene from the major methodological and biological motivation for this thesis, paralogs DBLMSP and DBLMSP2 (DBs). In Chapter 4, I study the DBs in greater detail, using a global dataset of >3,500 Pf genomes. Building a genome-graph-based pipeline, I recover variation inaccessible to single-reference based approaches (GATK), before uncovering new biology. Expressing each diverged DB haplotype as a mosaic of the others, I find widespread recombination in each gene, and also discover recent evidence of gene conversion between the two genes. In summary, this thesis provides both methodological advances into genome-graph based genotyping, and practical insights into the genome biology of an important human pathogen.

Vault RNA are small non-coding polymerase III transcribed RNA, which are a crucial com- ponent of a large ribonucleoprotein complex called Vault. Despite only 5% of vault RNA being associated with the Vault complex, little is known about their Vault-independent cellular roles. Here I show that vtRNA are processed by Dicer into functional small vault RNA (svtRNA) fragments that associate with Argonaute 2. Nascent RNA sequencing revealed the highest number of predicted svtRNA1-2 target genes among up-regulated genes, in AGO2 and Dicer Knock Down cells, compared to other svtRNA. svtRNA1-2 target sites were detected predominantly in the intronic regions. Genomic deletion of vtRNA1-2 results in the severe inhibition of proliferation and up-regulation of genes sig- nificantly associated with cell membrane physiology, including ERBB4 and cell adhesion. Finally, I show that svtRNA1-2 hybridises to target genes in a dual-region complemen- tarity and loop protrusion is required for intron targeting. These results show a novel small vault RNA1-2 with a role in gene expression regulation, possessing unique molecular features for targeting genes in introns leading to nascent RNA silencing.

Dorsal root ganglia (DRG) neurons have been well described for their role in driving both acute and chronic pain. With cell bodies clustered in the spinal column, these neurons project from the periphery into the dorsal horn. Subtypes convert a range of stimuli into electrical signals, and have been traditionally defined by size, histological markers, projection pathways, and electrophysiological phenotypes. With the advance of sequencing technologies, DRG transcriptomes have been examined through whole ganglia, single cell, and more recently, deep subpopulation sequencing in naïve lines. Here, we study the deep transcriptional profiles of multiple murine DRG populations in acute and chronic pain states while considering sex interactions. We have exploited currently available transgenics to label and purify subtypes after injury for high throughput sequencing. Using bulk tissue samples, we are able to circumvent the issues of low transcript coverage and drop-outs seen with single cell datasets, increasing our power to detect changes in gene expression. In one experiment, we have studied the molecular changes in five populations, classified by their functional and molecular signatures in naive states. These include general nociceptors, peptidergic nociceptors, non-peptidergic nociceptors, C-LTMRs, and Ad + AB-RA-LTMRs. We see both stereotyped and unique subtype signatures in injured states after nerve injury at both an acute and chronic timepoint. While all populations contribution to a general injury signature, subtype enrichment changes can also be seen, possibly driven by the loss of transected NP neurons by 4 weeks. Within populations, there is not a strong intersection of sex and injury, but sex differences in na\"ive states - particularly in Ad + AB-RA-LTMRs - still contribute to differences in injured neurons. Using an SNI model of neuropathic pain, ipsilateral DRG contain both injured and intact neurons. The contributions of these populations to pain have yet to be delineated. Mirroring the techniques above, we next used an Atf3creERT2 line to label and sort fluorescent neurons prior to RNA-seq. We have complemented this with a pilot snATAC-seq study, showing that neurons cluster by subtype in naive and injured states. We have curated this data into an accessible, online platform. Together, this contributes a resource to the pain community to probe subpopulation and sex-specific differences after nerve injury at multiple timepoints.

Chronic diseases place a huge burden on society and are the leading cause of death worldwide. Obesity, which is currently a worldwide epidemic, is a risk factor for chronic disease. Understanding drivers of food choices may help to address the escalating obesity problem. Many people are at an increased risk of obesity as they carry genetic variants linked to particular eating behaviours. Identification of causal variants can help to understand which genes and biological mechanisms underlie food preference. Previous work used UK Biobank data from ~500,000 individuals to identify 302 loci associated with food consumption. This research project aimed to understand which genes were responsible for these genetic associations, prioritise genes for functional validation study, and understand the function of these genes through in vivo food preference studies. It is well-known that there can be intake-related bias in food frequency questionnaire responses due to participants over- or under-reporting dietary intake and this makes nutritional research difficult. Mendelian randomization was used to statistically identify reporting bias in the UK Biobank food frequency questionnaire responses and these were subsequently corrected by the group using a bespoke method. A 6-stage gene prioritisation strategy was developed to identify a candidate gene for functional validation in vivo. Genes were allocated points based on three main components: gene expression in humans and mice, conservation of gene product across species using pairwise protein alignments, and knockout phenotype details from available mouse models. Genes were ranked in order of preference for functional follow-up study. The top two genes identified were IER5L (associated with meat and fat) and DCAF12 (associated with salt). It was difficult to determine a coherent role for IER5L in food preference. The top SNP mapped to IER5L was found to be in an eQTL containing two other genes: CRAT and PPP2R4. It was not possible to exclude any of these genes from the meat/fat association, but there were clear links in the literature between CRAT and food intake. CRAT was selected for functional follow up, along with the DCAF12 gene associated with adding salt to food. However, it was not possible to follow up the CRAT gene due to logistical reasons. The project instead progressed to the in vivo stage using the well-known MC4R appetite regulatory gene to develop a model of food preference that could be used with Crat-/- mice in the future. A novel 3-food preference model was developed using Mc4r-/- mice and WT mice. Mice were given a choice of three diets: protein-enriched diet (to model the CRAT meat association), fat-enriched diet (to model the CRAT fat association) and standard chow. Mice preferred to eat a completely fat diet or a mixed choice diet regardless of genotype. Mc4r-/- mice were heavier than WT mice but were not hyperphagic as was expected from previous studies. As expected, Mc4r-/- mice demonstrated hyperinsulinemia compared to WT mice but there were no differences in glucose or triglyceride levels found between Mc4r-/- and WT mice. In vivo modelling of the DCAF12 salt association was carried out using Dcaf12-/- and WT mice to investigate salt preference. Mice were given a choice of two bottles of water followed by a choice of 0.4% (75mM) salt solution and water. This was repeated using a higher concentration 0.8% (150mM) salt solution. Dcaf12-/- mice did not show a preference for the salt solution as expected, and preference reduced as salt concentration increased. Unexpectedly, mice demonstrated a right-hand side preference during the water bottle choice period. Furthermore, mice consumed less food as salt concentration increased. This salt-preference study highlighted a number of complexities that can be associated with translating human effects into in vivo models. Research into the genetics of eating behaviour can help to identify and understand the impact of genetic variants on dietary pathways. If we can understand biological drivers of individual food choices, this information may be used to guide people towards a healthier lifestyle and reduce levels of obesity.

Microorganisms shape the living world through activities encoded in their DNA sequences. Synthetic biology often involves modifying the behaviour of microorganisms with designed DNA made up of genetic parts to achieve complex biological functionality. Large libraries of genetic designs can be constructed using DNA assembly and amplification, which could enable a broader understanding of how to design sequences encoding specific functions. However, methods to characterise sequence and function are often limited in scalability and scope since they can only characterise individual genetic parts. DNA and RNA sequencing open new possibilities for characterising entire DNA libraries. Here, we apply nanopore sequencing, a technology where DNA sequences are read as they pass through nanoscale pores (nanopores), to study assembled libraries of genetic parts. DNA sequencing shows that one-pot combinatorial DNA assembly by ligation reliably constructs libraries from multiple genetic parts and that library composition is significantly more uniform when DNA is amplified without protein expression. The libraries encode transcriptional terminator genetic parts that signal where transcribing RNA polymerases (RNAPs) should dissociate from DNA. We use terminators as 'transcriptional valves' to tune RNAP read-through and control transcript isoform abundance, offering a new mechanism to regulate genetic designs transcriptionally. We develop a method to characterise the in vitro transcription of valve libraries simultaneously at nucleotide resolution using nanopore direct RNA sequencing (dRNA-seq). This method reveals that upstream sequence context changes how much termination occurs, along with genetic design principles for tuning and insulating terminator function. Using this knowledge, we then engineer an array of CRISPR guide RNAs transcriptionally regulated by our valves. With DNA being the code of life, synthetic biology comes with a responsibility to the living planet. Therefore, a framework for responsible innovation is used to assess potential real-world impacts of the arrays. This work provides new avenues for studying DNA library composition, regulating transcription and innovating biotechnology responsibly, demonstrating the value of sequencing for exploring complex sequence-function landscapes.

Alongside tremendous challenges in infectious diseases, like the rise of antimicrobial resistance and the coronavirus disease pandemic, the 21st century is also witness to the big data revolution, which offers opportunities to design methodology capable of addressing these great challenges. Whilst developing tools there are two competing philosophies of how to gain insight from big data: The modelling approach, where the natural data generating mechanism is approximated by statistical inference, and the algorithmic approach, where general-purpose algorithms are tuned to capture hidden structure in the data for prediction. The aim of the thesis is to contribute existing infectious disease problems, by motivating, designing, and applying the correct big data methodology, whilst facilitating future use through generating applications can be easily re-purposed. I first design a machine learner that can predict the source of Campylobacteriosis 33% more accurately than the previous most commonly used methods. Our method broadens the data input spectrum to captures of whole genomes, which uniquely allows assigning sources to individual samples showing a shift in host affinity of one of the most common lineages of Campylobacter jejuni. Based on the individual prediction of the machine learner,I infer which genetic changes are associated with host specificity by conducting a genome-wide association study. I find fluoroquinolone resistant genes pre-adapting chicken isolates to infection for humans and polyphosphate pathway associated genes to distinguish adaption to chicken and ruminant niche. For the study of COVID-19 risk, I conduct a machine learning prediction of very severe forms of the disease, hospitalisation, and susceptibility, whilst also inferring risk factors for all phenotypes by applying Bayesian model averaging. I re-discover commonly defined risk factors describing socio-economic standing, ill health and ethnicity whilst discovering more novel factors like previous lung injury predisposing very severe COVID-19 and bring order to the wealth of published COVID-19 risk studies. In the closing arguments I give limitations of my work and give recommendations on how the developed tools can be re-applied to make big data research more accessible. I also expand how statistical inference and machine learning prediction can be used in unison to tap into the potential of big data to address the foremost infectious disease challenges of our time.

Multiple myeloma (MM) is an incurable cancer of plasma cells, with an average five-year survival rate of approximately 50%. Over the last two decades, application of MM therapeutics, namely proteasome inhibitors (PI) and immunomodulatory imide drugs (IMiD), have almost doubled median survival time of MM patients. However, most patients will develop drug resistance. Acquired anti-cancer drug resistance remains one of the biggest barriers in the treatment of myeloma. Therefore, identifying novel therapeutics effective against MM, which are capable of overcoming drug resistance is of the utmost importance. Recently, the prolyl-tRNA synthetase (ProRS) inhibitor, halofuginone (HF), has been shown to be effective against cancer, including one study demonstrat- ing effectiveness against MM. Halofuginone is an ATP-dependent, proline and tRNA competitive inhibitor of ProRS. Recently, a new proline-uncompetitive, ATP-competitive ProRS inhibitor, NCP26, has been developed. In this work the application of HF and NCP26 in MM is explored using both MM cell lines and primary MM patient samples. HF and NCP26 demonstrated cytotoxicity to both drug-sensitive and proteasome inhibitor (PI)-resistant MM cell lines in a dose dependent manner. The anti-MM effects of HF were abrogated in the presence of excess proline, whereas excess proline had no effect on NCP26 treatment. Using bulk RNA-seq, HF and NCP26 treatment was shown to activate the amino acid starvation response (AAR), endoplasmic reticulum (ER) stress and downstream apoptotic signalling pathways in MM cell lines. NCP26 resistant MM cells also showed significant suppression of ER-associated elements, as well as reduced expression of ATF4 (the master regulator of amino acid metabolism). Single-cell RNA- seq (scRNA-seq) was employed to further examine the effects of ProRS inhibition in primary patient bone marrow samples. scRNA-seq confirmed the activation of the AAR, ER stress and apoptotic mechanisms in newly-diagnosed and relapsed MM patient cells. From differential expression and composition analyses, HF and NCP26 treatment was shown to selectively target MM cells over the majority of immune cells, including NK cells, B cells and T cells. However, HF and NCP26 were also shown to elicit apoptosis in myeloid cells. From this pre-clinical validation of ProRS inhibitors, it is clear that ProRS represents a new, promising target in MM. A number of computational methods were developed to support the experiments conducted in this work. scRNA-seq and tRNA-seq analysis pipelines were developed and their performances were benchmarked. An MM classifier was also built to be able to automate cell type classification of MM patient scRNA-seq samples. The whole bone marrow (WBM) MM scRNA-seq data from this work were used to train the MM classifier. Publically available MM WBM scRNA-seq datasets were used to evaluate the performance of the MM classifier, and it was found to be fairly accurate at labelling MM cells.

Endemic and emerging viruses are essential to study in order to understand their prevalence and disease association within Northern Ireland's livestock and wildlife as, they may be considered an important risk factor to food security, food safety and public health. In recent decades, the prevalence of viral zoonotic pathogens transmitted from animals to humans is currently creating an increasing socio-economic impact. Accordingly, routine pathogen molecular epidemiology surveillance is the key to the identification and mitigation of new and endemic viruses but is often lacking with undiagnosed diseases not being thoroughly investigated. This thesis focuses on four research areas that fall within the scope of molecular detection and investigation of emerging and endemic viruses in Northern Ireland's livestock and wildlife including; the study of novel porcine parvoviruses (PPV) coinfection status with porcine circovirus 2 (PCV2) and statistical disease association, hepatitis E virus (HEV) prevalence, enteric virome of clinically diseased animals and the prevalence and genotyping of rotavirus (RV) by utilising real-time PCR, Sanger sequencing, Illumina second generation sequencing and bioinformatic analysis. The recently identified novel PPVs 2-4's clinical significance is poorly understood. Pig endemic diseases for example porcine respiratory disease complex and porcine circovirus associated disease are caused initially by primary viral infections, however disease association by coinfections with novel parvoviruses and porcine circovirus 2 remains to be determined. The analysis of archival samples from clinically diseased pigs' indicates that PPV2 is an emerging pathogenic agent, and is statistically associated with porcine respiratory disease complex in pigs. The emerging HEV is ubiquitous in pig herds worldwide, and HEV genotypes 3 and 4 are zoonotic viruses transmitted mainly as a foodborne disease leading to subclinical to acute infections in humans. A HEV molecular epidemiology study on the prevalence of HEV through an abattoir study and the dynamics of the HEV infection in a longitudinal study were completed. The enteric microbiome is a complex community of bacteria, viruses, fungi and protozoa. This a new area of research where limited studies exists on the enteric virome of livestock and wildlife. Studies of the virome will aid in the understanding of; viral taxonomic identification, viral prevalence, virus genetic and phenotypic diversity, coinfections and virus-bacteriome and host interactions modelling. Next generation viral metagenomics is a new tool for monitoring and detecting new emerging viruses and will help to fill the knowledge gap of the mammalian virome composition. Rotavirus group A (RVA) is the major cause of acute diarrhoea and severe dehydration in children and young mammalian host species. Infection of neonate and young animals may lead to death, resulting in economic losses for the farmer. A rotavirus molecular epidemiological surveillance study was conducted mainly on clinically diseased Northern Ireland animals by Sanger sequencing for RVA detection, NGS whole genome amplification and phylogenetic analysis. The study revealed a high level of rotavirus group A phylogenetic diversity and demonstrates the importance of accumulating knowledge of disease and molecular epidemiological surveillance coupled with viral metagenomics for understanding the viruses circulating in Northern Ireland livestock and wildlife.

Genetics contribute substantially to the ability of the immune system to respond appropriately to a challenge. Consequently, many infectious and inflammatory diseases have a heritable component. As genome-wide association studies provide increasing data linking genetic variants to disease, we can leverage this information to gain insights into disease biology if we can elucidate the mechanisms underlying an observed association. Data-driven bioinformatic approaches, in vitro mechanistic studies and whole-organism approaches to study integrated pathophysiological systems provide complementary information to help establish causal links between specific variants and their effects on a target gene, and between that gene and disease pathogenesis. Cells of the mononuclear phagocyte system play key roles in regulation of innate and adaptive immune responses, either via secreted mediators or through direct cell-to-cell contacts. To advance understanding of how these regulatory processes, and genetic variations therein, shape the course of disease, I have combined experimental and bioinformatic approaches to explore novel genetic associations between genes for macrophage surface receptors and human disease. Specifically, functional follow-up of an association between adhesion G-protein-coupled receptor CD97 and severe influenza showed, using a mouse model, that deficiency of this receptor reduces the efficiency of the CD8+ T-lymphocyte response, a process critical to clearance of infected cells. Secondly, I addressed the question of how to identify causal variants in a disease-associated linkage disequilibrium block, for an association between macrophage regulatory receptor SIRPα and schizophrenia. To achieve this, I developed novel methodology for targeted locus screening using CRISPR/Cas9 mutagenesis, and identified a number of plausible causal regulatory variants that could affect expression of this receptor. Combining variant-level information with gene-level studies of disease pathophysiology can provide valuable insights into genetic causation of immune dys-regulation leading to disease, which may be harnessed for improved personalised disease risk prediction, or to identify therapeutically targetable pathways.

The Global Burden of Disease Study has shown that kidney disease is one of the fastest rising causes of mortality worldwide and claims 1.1 million lives per year. A range of factors can cause kidney damage. The kidneys filter approximately 180 litres of blood per day, and a decline in kidney function is quickly followed by a severe decline in health and eventually death. A late-stage indicator of kidney disease is proteinuria, and a common feature connecting many kidney diseases is damage to podocyte cells. Podocyte cells are a vital cell type involved in blood filtration and the maintenance of kidney function. Podocytes are highly specialised cells that envelope the glomerular capillaries and form part of a selective barrier, ensuring the retention of larger proteins within the lumen of the capillaries they encapsulate. Increasing the understanding of podocyte cell physiology will allow us to gain novel insight, improve our knowledge of cellular interactions within podocytes, and hopefully lead to novel therapeutic techniques. Elucidating the physiology of the kidney will enable the development of improved therapies and treatments for managing kidney diseases. In this study, novel transgenic zebrafish podocyte reporter lines have been created as informative tools in analysing podocyte cell function and changes within podocytes. The rapid external development of larvae and the optical clarity of zebrafish made it an ideal model for this project, where observation and analysis of podocyte cells in vivo and the ability to image the transparent larvae was an essential aspect of this work. In this study, using a range of advanced imaging techniques and the analysis of the reporter lines, it has been possible to capture changes to podocyte cells. The overarching aim of this work was to gain greater insight into podocyte physiology. Using gateway cloning, informative, podocyte-specific, transgenic zebrafish lines were generated and used as tools to study podocyte cells. These lines used podocyte-specific fluorescent proteins to identify and locate podocytes within larvae and adults, allowing these cells to be targeted, isolated, and monitored in experimental use. Using SPIM (Single/ Selective Plane Illumination Microscopy), it was possible to induce injury within podocytes of larval fish and visualise changes in calcium ion dynamics in real-time. Through the microinjection of the known nephrotoxic agent Puromycin aminonucleoside, it was possible to induce oedema in larval fish and assess the morphological changes to podocyte cells through analysis via transmission electron microscopy. Using the novel GCaMP6s LifeActTagRFP-T zebrafish line the structure of the actin cytoskeleton was assessed in closer detail. Bioinformatics approaches were used to develop a pathway diagram elucidating the interactions within podocyte cells through manual curation of literature, the use of current datasets, and incorporating novel scRNA seq data. This detailed diagram presents an interactive network that allows the visualisation of podocyte cell interactions in a way that is more accessible and permits visualisation of the data in ways that would not be possible with graphs and images alone. It allows for the identification of discrepancies in metabolic pathways and intracellular communications to be displayed; it also contributes to what is currently known about podocyte physiology. Finally, using GCaMP6s and LifeActTagRFP-T tagged podocyte cells disassociated from adult fish, it was possible to generate a single-cell RNA sequencing (scRNA-seq) library, detailing active transcripts present within healthy zebrafish podocytes and the identification of distinct cell populations and podocyte cell markers, adding to and allowing comparisons with libraries generated from zebrafish podocytes in alternative states, such as having undergone injury, or in alternative species. The combination of these techniques and outcomes of this work has allowed the presentation of a project which has contributed innovative transgenic zebrafish reporter lines, the novel display of an increase of calcium ion concentration in injured podocyte cells, the production of a detailed, novel scRNA seq library and a pathway diagram utilising bioinformatics techniques, that is interactive and informative. These contributions can add to the generation of new therapies for treating cardiovascular diseases and have provided additional transgenic zebrafish lines to be used as a model to study kidney disease development, injury, and repair. The Global Burden of Disease Study has shown that kidney disease is one of the fastest rising causes of mortality worldwide and claims 1.1 million lives per year. A range of factors can cause kidney damage. The kidneys filter approximately 180 litres of blood per day, and a decline in kidney function is quickly followed by a severe decline in health and eventually death. A late-stage indicator of kidney disease is proteinuria, and a common feature connecting many kidney diseases is damage to podocyte cells. Podocyte cells are a vital cell type involved in blood filtration and the maintenance of kidney function. Podocytes are highly specialised cells that envelope the glomerular capillaries and form part of a selective barrier, ensuring the retention of larger proteins within the lumen of the capillaries they encapsulate. Increasing the understanding of podocyte cell physiology will allow us to gain novel insight, improve our knowledge of cellular interactions within podocytes, and hopefully lead to novel therapeutic techniques. Elucidating the physiology of the kidney will enable the development of improved therapies and treatments for managing kidney diseases. In this study, novel transgenic zebrafish podocyte reporter lines have been created as informative tools in analysing podocyte cell function and changes within podocytes. The rapid external development of larvae and the optical clarity of zebrafish made it an ideal model for this project, where observation and analysis of podocyte cells in vivo and the ability to image the transparent larvae was an essential aspect of this work. In this study, using a range of advanced imaging techniques and the analysis of the reporter lines, it has been possible to capture changes to podocyte cells. The overarching aim of this work was to gain greater insight into podocyte physiology. Using gateway cloning, informative, podocyte-specific, transgenic zebrafish lines were generated and used as tools to study podocyte cells. These lines used podocyte-specific fluorescent proteins to identify and locate podocytes within larvae and adults, allowing these cells to be targeted, isolated, and monitored in experimental use. Using SPIM (Single/ Selective Plane Illumination Microscopy), it was possible to induce injury within podocytes of larval fish and visualise changes in calcium ion dynamics in real-time. Through the microinjection of the known nephrotoxic agent Puromycin aminonucleoside, it was possible to induce oedema in larval fish and assess the morphological changes to podocyte cells through analysis via transmission electron microscopy. Using the novel GCaMP6s LifeActTagRFP-T zebrafish line the structure of the actin cytoskeleton was assessed in closer detail. Bioinformatics approaches were used to develop a pathway diagram elucidating the interactions within podocyte cells through manual curation of literature, the use of current datasets, and incorporating novel scRNA seq data. This detailed diagram presents an interactive network that allows the visualisation of podocyte cell interactions in a way that is more accessible and permits visualisation of the data in ways that would not be possible with graphs and images alone. It allows for the identification of discrepancies in metabolic pathways and intracellular communications to be displayed; it also contributes to what is currently known about podocyte physiology. Finally, using GCaMP6s and LifeActTagRFP-T tagged podocyte cells disassociated from adult fish, it was possible to generate a single-cell RNA sequencing (scRNA-seq) library, detailing active transcripts present within healthy zebrafish podocytes and the identification of distinct cell populations and podocyte cell markers, adding to and allowing comparisons with libraries generated from zebrafish podocytes in alternative states, such as having undergone injury, or in alternative species. The combination of these techniques and outcomes of this work has allowed the presentation of a project which has contributed innovative transgenic zebrafish reporter lines, the novel display of an increase of calcium ion concentration in injured podocyte cells, the production of a detailed, novel scRNA seq library and a pathway diagram utilising bioinformatics techniques, that is interactive and informative. These contributions can add to the generation of new therapies for treating cardiovascular diseases and have provided additional transgenic zebrafish lines to be used as a model to study kidney disease development, injury, and repair.

Whole genome sequencing (WGS) is increasingly used to diagnose rare genetic diseases (RGD). WGS data contains approximately 5,000,000 variants per patient, out of which often one variant causes disease. Variant prioritisation algorithms (VPA) help identify disease-causing variants and functional studies can provide evidence to confirm a variant's effect on a gene or gene product. In this thesis, I test and develop VPA on simulated and real patient data and conduct a functional study to contribute to the validation of HDLBP as a novel disease gene for Fine-Lubinsky syndrome (FLS). Two established analysis frameworks, Exomiser and VAAST+Phevor, contain VPA that only use genotypic data (GA) and VPA that use genotypic and phenotypic data (GPA). The GA and GPA performance is benchmarked on eleven real WGS patient cases for which disease-causing variants were previously identified in known and novel genes. The GPA performed better than the GA, ranking more benchmark variants first (Exomiser: 4 vs. 5; VAAST+Phevor: 1 vs. 8), whilst reducing the percentage of variants requiring further analysis (Exomiser: from 32.3% to 2.2%; VAAST+Phevor: from 25.4% to 11.2%). Identifying disease-causing variants in genes that are novel for a specific phenotype is challenging for GPA. A VPA called GPET is developed for disease-causing variant identification in novel genes based on genotypic, phenotypic, and tissue-specific expression data. GPET is benchmarked against Exomiser on a simulated dataset of disease-causing and non-disease-causing variants with imperfect phenotypic annotations, achieving an Area Under the Curve of 0.95, compared to 0.91 for Exomiser. GPET performs worse than Exomiser on the eleven RGD cases for known disease genes and better for all novel disease genes without phenotypic annotations. Patients for one of the RGD cases are affected by FLS and carry a candidate variant in HDLBP. The variant is suspected to cause FLS by decreasing the RNA-binding activity of vigilin, the protein encoded by HDLBP. A functional study based on oligo(dT) capture is conducted, confirming reduced RNA-binding activity of vigilin due to the candidate variant. The investigations in this thesis demonstrate that phenotypic and tissue-specific expression data can improve VPA performance for novel disease genes and provide evidence to support HDLBP as a disease gene for FLS.

Non-target site resistance (NTSR) to herbicides in weeds is a threat to global crop yields. Few specific alleles conferring NTSR have been identified, only two of which involve loss of gene function. Here, I show that the mechanism of NTSR to the herbicide thaxtomin A (TXTA) conferred by loss-of-function of the gene PAM16 is conserved in Marchantia polymorpha, validating its use as a species with which to study NTSR. To identify novel NTSR mechanisms, I used forward genetics in M. polymorpha to generate 13 TXTA-resistant mutants and identified candidate NTSR-conferring SNPs in these mutants via mapping-by-sequencing. I identified a novel mechanism of NTSR to TXTA conferred by loss-of-function of the gene RAD8. To functionally characterise this mechanism, I generated rad8 loss-of-function mutants and found that they are cross-resistant to isoxaben, overproduce reactive oxygen species, and produce less of a putative TXTA metabolite than wild-type. I also found 4 candidate resistance-conferring SNPs in M. polymorpha homologues of genes associated with NTSR in the weed Amaranthus tuberculatus. These results demonstrate that forward genetics in M. polymorpha can identify novel NTSR alleles and corroborate existing candidate NTSR alleles, and that reverse genetics in M. polymorpha can be used to functionally characterise these resistance mechanisms.

Colorectal cancer (CRC) is a multifaceted and heterogeneous disease. The molecular profiling of CRC has made it possible to detect tumours with unique biological features. Gene expression-based signatures have been at the forefront of cancer research in providing valuable prognostic and predictive information. However, these published signatures rarely make any clinical impact. This thesis utilises in silico methods to interrogate a pre-existing gene signature and evaluate the underlying CRC biology defined by the stratification based on the gene signatures. Additionally, accessing the published gene signatures, a new classification system was developed that defines pathway-driven tumours for patient stratification and tested its prognostic value. A transcriptional breast cancer (BC) originated DNA Damage Immune Response (DDIR) assay, revealed overlapping proinflammatory-like tumour biology in CRC, but with no clinical benefit from oxaliplatin (Chapter 3). Utilising transcriptional gene signatures from the gene set database, pathway-level unsupervised class discovery and classifier development from KRAS mutant CRC revealed three groups, named KM1, KM2 and KM3 (Chapter 4). The biological and prognostic evaluation revealed that KM1 were MYC driven cell-cycle activated, tubular, LGR5+ stem-like tumours with good prognosis, KM2 were stromal/immune rich, of serrated origin and regenerative stem-like biology with diverse pathway activation, including TGF-β activation, EMT and immune responses, and an intermediate prognosis. A novel finding of this classification was KM3, with large repression of pathways such as cell cycle, EMT, KRAS signalling activity, but elevation in neurotransmitter- and enteroendocrine-like biology, highly differentiated tumours, the potential inclusion of epigenetic dysregulation in maintaining stem plasticity, and a clinically worse prognosis (Chapter 5). In summary, this thesis provides a new pathway-driven classifier for CRC that includes a novel, previously uncharacterised subset of CRC subtype as KM3 with clinically poor outcome, which provides a unique biological phenotype that needs further research to translate into the clinic.

Hereditary peripheral neuropathy, also known as Charcot-Marie-Tooth disease (CMT) and related disorders, are a group of genetic disorders causing length-dependant neuropathy, resulting in motor and/or sensory loss progressing from the lower extremities toward the spine. With a population prevalence of 1 in 2,500, CMT is the most common inherited neuromuscular disorder and among the most common of all Mendelian diseases. Despite advances in the understanding of CMT genomics, a significant number of CMT patients remain genetically undiagnosed, and disease-modifying therapeutics are not available for the vast majority of CMT patients. In this thesis I present four topics addressing related areas of unmet need within the CMT field. My aim is to identify biological features relevant to a broad range of subtypes and experimental models, bridging the division between clinical and basic science. Firstly, I perform a systemic review of therapeutic studies of inherited peripheral neuropathies, with a focus specifically on gene and variant-level, rather than disease-level, treatment outcomes. Therapeutics for neuromuscular diseases are often gene- or mutation-specific, and I identify several variant-specific therapies and therapeutic approaches for inherited neuropathies, as well as variant-specific outcomes for gene-specific therapies, to be integrated into an online genetic diagnostic platform. Within the clinical trials I identify lack of responsive CMT biomarkers as a major limiting factor for therapeutic trials. Therefore, I next seek to identify novel biomarkers, by applying proteomics in a cohort of CMT patients and mouse models to identify neural cell adhesion molecular 1 (NCAM1) and growth differentiation factor 15 (GDF15) as serum biomarkers of CMT. NCAM1 may potentially function as a disease-progression biomarker, whereas GDF15 is indicated to be a very accurate diagnostic biomarker. Additionally, I find that broad elevation of serum complement proteins across CMT subtypes, which, while less promising as biomarkers, may indicate that dysregulation of this system. Thirdly, whilst the pathophysiology of some individual CMT genotypes is well understood, in many cases there is not currently a clear shared mechanism linking basic CMT-associated gene families to neuropathy pathogenesis. In particular, it is unclear why dominant missense mutations in several tRNA synthetases lead to axonal neuropathy, whilst other variants in these genes cause very variable human disease. Therefore, I develop an induced neural progenitor cell (iNPC) model derived from skin fibroblasts of patients with these genetic defects and perform transcriptomic and proteomic analyses of dysregulated pathways. Finally, to address the particularly large proportion of undiagnosed patients with distal hereditary motor neuropathy, I test the possibility of using functional transcriptomic and proteomic datasets to identify causative variants of axonal neuropathies, and develop a machine learning model to identify neuropathy-causative variants within large datasets of whole exome or genome sequencing. Although CMT is highly heterogenous at a genetic level, I show across these topics that there are several molecular signatures common across many subtypes. The identification of two non-subtype-specific biomarkers will aid the development of CMT therapeutics. Dysregulation of cell adhesion molecules is identified as a major feature in both patient sera and patient-derived iNPC models, while the machine learning model I develop demonstrates that protein and RNA enrichment in the axon is a key feature of CMT.

Cyanobacteria have lived on Earth for longer than animals, plants, and fungi. In that time, they have oxygenated the atmosphere, rendering the planet habitable for complex life. Yet whilst the antiquity and diversity of Cyanobacteria is appreciated, the timing and evolution of genes guiding their interaction with the world are not well understood. In this thesis, sequencing technology, molecular clocks, and phylogenetic analyses are employed to search for evidence of ancient genes managing oxidative stress and their transition from unicellular to multicellular morphology. This is accomplished by interrogating hundreds of Cyanobacterial genomes for three antioxidant enzymes and eleven multicellularity genes. The evolutionary history of those genes is then reconstructed and mapped onto Bayesian molecular clocks calibrated with Precambrian microfossils, to predict when they first appeared in ancestral populations. These analyses suggest that stem Cyanobacteria appeared approximately a billion years before the Great Oxidation Event and provide the first phylogenetic evidence that oxygenic phototrophs used CuZnSOD to manage oxidative stress before oxygen accumulated on a global scale. In this largely anoxic, microbial-dominated period of Earth's history, genes underlying multicellular phenotypes emerged in stages. The first encoded a transcriptional regulator, permease, and components of septal junctions. Approximately 75 to 200 million years later, during the Great Oxidation Event, two further genes managing the frequency of differentiated and vegetative cells in filaments evolved. By sequencing five draft genomes, brackish Cyanobacteria from a coastal beach are also found genetically capable of synthesising glucosylglycerol to handle salinity in a similar manner to marine species. Altogether, this research expands a growing body of evidence describing how photosynthetic bacteria have adapted to their changing physical environments over the last 3.6 billion years. It also highlights the wealth of information in public data repositories, which can be analysed to better understand the co-evolution of life and Earth.

Sex is arguably the most important differentiating characteristic in most mammalian species, separating populations into different groups, with varying behaviors, morphologies, and physiologies based on their complement of sex chromosomes, amongst other factors. In humans, despite males and females sharing nearly identical genomes, there are differences between the sexes in complex traits and in the risk of a wide array of diseases. Sex provides the genome with a distinct hormonal milieu, differential gene expression, and environmental pressures arising from gender societal roles. This thus poses the possibility of observing gene by sex (GxS) interactions between the sexes that may contribute to some of the phenotypic differences observed. In recent years, there has been growing evidence of GxS, with common genetic variation presenting different effects on males and females. These studies have however been limited in regards to the number of traits studied and/or statistical power. Understanding sex differences in genetic architecture is of great importance as this could lead to improved understanding of potential differences in underlying biological pathways and disease etiology between the sexes and in turn help inform personalised treatments and precision medicine. In this thesis we provide insights into both the scope and mechanism of GxS across the genome of circa 450,000 individuals of European ancestry and 530 complex traits in the UK Biobank. We found small yet widespread differences in genetic architecture across traits through the calculation of sex-specific heritability, genetic correlations, and sex-stratified genome-wide association studies (GWAS). We further investigated whether sex-agnostic (non-stratified) efforts could potentially be missing information of interest, including sex-specific trait-relevant loci and increased phenotype prediction accuracies. Finally, we studied the potential functional role of sex differences in genetic architecture through sex biased expression quantitative trait loci (eQTL) and gene-level analyses. Overall, this study marks a broad examination of the genetics of sex differences. Our findings parallel previous reports, suggesting the presence of sexual genetic heterogeneity across complex traits of generally modest magnitude. Furthermore, our results suggest the need to consider sex-stratified analyses in future studies in order to shed light into possible sex-specific molecular mechanisms.

The response of the global genomics community to the SARS-CoV-2 pandemic has been unprecedented. At time of writing there are more than 3.7 million SARS-CoV-2 genome sequences shared publicly on GISAID (www.gisaid.org). This scale of data on that order of magnitude presents novel opportunities and challenges for the field of genomic epidemiology. This thesis describes the development, validation and implementation of novel tools to facilitate different aspects of genomic epidemiology, from outbreak investigations to surveillance efforts. The Pango nomenclature lineage system is a set of rules that defines epidemiological lineages of SARS-CoV-2. Pango defines lineages from whole genome sequences, which 195 nations around the world have been producing for SARS-CoV-2. In chapter 1, I discuss the development and validation of pangolin, a software tool developed to assign the most likely Pango lineage to novel SARS-CoV-2 genomes. Initially, pangolin used a classic phylogenetic approach to assign lineages although further methods were trialled and implemented as the pandemic progressed to cope with the scale of and analytical challenges associated with SARS-CoV-2 data. Since it was first implemented, millions of SARSCoV-2 genomes have been assigned lineages with the pangolin tool from users across the world. For a number of reasons, labs may not be in a position to produce full genome sequences. Chapter 2 investigates how the lineage system can be used if only spike nucleotide sequences are available and defines 'lineage sets' that summarise what lineage information exists within a given spike haplotype. We find that for many lineages, including the main lineages corresponding to the WHO-defined variants of concern (VOCs), the spike nucleotide sequence is sufficient to distinguish Pango lineages and I describe the development of a software tool hedgehog that is a wrapper for pangolin that both defines and assigns these spike-based lineage sets. Pango lineage assignments with pangolin have been used almost ubiquitously across the globe and provide a simple, quick piece of information to classify SARS-CoV-2 genomes. However, for both outbreak investigations and routine surveillance, a more in-depth analysis is needed to give more than just this one piece of information. In chapter 3, I present civet, a software tool that addresses the challenge of the SARS-Cov-2 global dataset that is on the order of 3.7 million sequences and performs robust phylogenetic analyses on query sequences of interest, whilst contextualising them in the background data. Using civet, the user can produce an interactive report that summarises genomic, phylogenetic and epidemiological information, enabling routine analyses and investigations to be carried out in a single command. The suite of tools in this thesis have been developed to enable researchers to rapidly get robust and actionable information from SARS-CoV-2 genomes for genomic epidemiology efforts worldwide.

The high number of available reference genomes for different species and their comparison has enabled the elucidation of gene birth mechanisms that act over a long evolutionary timescale. However, the lack of several reference-quality genomes for different individuals of the same species has hampered the study of the mechanisms of more evolutionarily young gene births. Despite the high throughput brought about by second-generation sequencing technologies, their short read length has limited the study of genetic diversity to single nucleotide polymorphisms (SNPs) and short indels. However, in order to study gene-level events, we need to characterise the genetic diversity of a species comprehensively, including structural variants (SVs) (> 50 bp). I present the most comprehensive set of genomes and SVs for Caenorhabditis elegans. I have assembled a high-quality genome for each of 20 wild isolates of the nematode using long and short read sequencing. I show that 1,587 transcripts are deleted among the wild isolates and thus sketch the  first definition of the core genome of C. elegans. I present the case of a highly proliferative transposon harbouring a transcription factor binding site (TFBS) and use it to address the question of transposon co-option in this model organism. Finally, using this dataset, I show that tandem gene duplication is a prominent gene birth mechanism, whereas horizontal gene transfer (HGT) played little or no role in the birth of recent C. elegans genes. Additionally, I show that G protein-coupled receptors (GPCRs) have high levels of presence/absence variation (PAV) and discuss the significance of this  finding in light of the ecology of this little worm.

Rare monogenic diseases affect around 300 millions worldwide at any given time. Although genetically and clinically heterogenous, rare diseases are frequently caused by pathogenic mutation in a single gene; in particular, missense and in-frame insertion and/or deletion (indel) variants. Differentiating the few truly pathogenic alterations from the background of functionally neutral variants is a challenge which has resulted in a translational bottleneck. Largely due to the absence of segregation and functional data, the majority of reported variants-of-interest remain clinically defined as variants of unknown significance (VUS). Since experimental validation of all variants is impractical, computational prediction tools have been designed to attempt to prioritise variants for downstream analysis. While accurate variant prioritisation can lead to fewer VUS and a higher rate of diagnosis, this has yet to reach clinical standard. The aim of this study was to i) design a protein-specific prediction model to improve missense variant prioritisation and ii) use integrative structural analysis to differentiate pathogenic from benign in-frame indel variants. Using a protein-specific approach to variant analysis, structural and clinical data were combined using machine learning to differentiate pathogenic from benign missense variants in a subset of genes. The resulting prediction model, ProSper (protein-specific variant interpreter), outperformed a number of widely-used or recently developed missense prediction tools for the majority of genes studied. For a group of in-frame indel variants that were classified as VUS in a lab, structural interpretation was used to show variant clustering and accurately differentiate pathogenic variants in a family of genes. Variant prioritisation and interpretation using these approaches in clinically relevant situations can improve rare disease diagnosis and patient management. In addition, structural analysis at the atomic-level can be used to assess a variant's structural impact and provide data on the molecular mechanism of disease. This research adds to the body of evidence showing improved accuracy in variant prediction and prioritisation in groups of genes using a protein-specific approach compared to general predictors. Incorporating integrative structural analysis into the diagnostic framework can accelerate variant interpretation and minimise the rate of VUS leading to correct and timely diagnosis in more patients.

Cardiac conduction system (CCS) morphogenesis is essential for correct heart function yet is incompletely understood. During my D.Phil, I established the transcriptional landscape of cell types populating the developing heart by integrating single-cell RNA sequencing and high-resolution spatially resolved transcriptomics (SrT) analyses. This led to the identification of a previously unrecognized cardiomyocyte population expressing dopamine beta-hydroxylase (Dbhh⁺ -CMs), which is closely associated with the CCS. To further comfirm, genetic fate mapping using DbhCre/Rosa26-tdTomato reporters combined with SrT revealed that Dbh⁺-CM lineage is Hcn4-independent, first emerged in the sinoatrial node at E12.5, then to atrial and ventricular CCS components at E14.5 and became prominent within CCS at perinatal stages. Dbhh⁺ -CM populating CCS occurred concurrently with sympathetic innervation. Electrophysiological optogenetic mapping revealed a fast conduction in DbhCre/ChR2-tdTomato hearts by selectively stimulating Dbh+ -CMs. Furthermore, catecholaminergic-type vesicles were identified in adult cardiomyocytes. Thus, by utilising advanced transcriptomic, genetic and functional analyses, my study provides new insights into mammalian CCS development and heterogeneity. Besides, by combining an improved heart tissueclearing technique with high resolution light sheet fluorescence microscopy (LSFM), I conducted three-dimensional and multi-scale volumetric imaging of ultra-thin planes of murine hearts to visualize the specific cardiomyocyte subpopulations. This method provides a tool for further geometrical and topological analysis of these cell types and the correlation with their physiological function in specific region of the murine heart.

Standard parametric statistical approaches based on comparison to global activity tend to perform poorly when this activity varies over multiple scales. Such multiscale variation, termed long range dependence, is a well-documented features of many biological and neurological data sets. We provide evidence from the literature as well as from data that demonstrates long range dependence across three contexts in: protein, brain and neuronal data. We propose novel non-parametric statistical approaches that account for these dependencies using network models. In networks of annotated proteins (nodes) connected by their physical interactions (edges), communities of functionally related proteins can provide possible novel drug targets. However, this guilt by association approach is vulnerable to potential bias in protein annotations. By restricted random-walk-based ranking of communities, our method, CommFinedWalker, preserves protein co-expression within well-ranked communities (suggesting functional conservation). The method provides novel drug target candidates that add to the diversity of interesting proteins and potential drug targets for research. As proof of concept, we explore a case study of a community that contains proteins involved in a rare genetic disorder. These results also serve as inspiration for novel methods to rank interesting communities in other contexts including brain state networks. Brain states have previously been studied across many conditions, including sleep and disorders of consciousness using a variety of models. However, brain state has not yet been studied in pharmacologically-induced unconsciousness. We define a new subclass of models, Hidden Markov Graph Models, to study brain states and communities of brain regions in fMRI data from subjects in both wakefulness and anaesthesia-induced unconsciousness. The framework draws on principles of free energy minimisation and entropy maximisation. We show evidence of characteristic resting state network activity in wakefulness as well as states of reduced temporal and functional complexity that are unique to anaesthesia-induced unconsciousness. These states further relate to established EEG markers of unconsciousness and long range dependent slow waves with implications for effective anaesthesia concentration in surgery. Lastly, we propose a pipeline to guide experimentation and perform automated detection of synaptic calcium events from single neuron light sheet microscopy data. This pipeline iteratively combines experimentation with experimentally informed simulation to improve experimental outcomes. The model fitting framework we provide draws on the free energy principle, and detection and smoothing methods that account for long range dependence and non-stationary activity present in the microscopy background. Through simulated results we empirically show that our methods can provide consistent estimates of actual experimental conditions that could be used to inform future experimentation.

Disruption to normal circadian rhythms and frequent periods of sleep deprivation (SD), are endemic in today's 24 hour society; therefore understanding the fundamental biology underlying what happens in the suprachiasmatic nuclei (SCN) during entrainment, and other specific brain regions following SD has never been more important. Current understanding of entrainment and SD within the SCN at the molecular level is still very limited, despite SD already having been shown to have a profound effect on brain function. The published Assay for Transposase-Accessible Chromatin sequencing (ATAC-seq) protocol was successfully modified to allow the technique to be effective with small amounts of tissue collected from murine SCN tissue. ATAC-seq was effectively used to investigate chromatin accessibility within cells of the SCN following light pulsing at different times of the day, and following SD in the SCN, dentate gyrus, and cortex. Data from ATAC-seq light pulse experiments was compared to published gene expression data following light pulsing and determined changes in chromatin accessibility also affect the transcriptome. RNA-seq was used to investigate gene expression following SD, and gain insights into how chromatin affects the transcriptome following SD. Results obtained indicate, chromatin accessibility is highly dynamic within all brain regions investigated. ATAC-seq revealed chromatin accessibility changes in the SCN following light pulsing, but only at certain times of the day, and discovered chromatin modification across all brain regions following SD. Several genes identified close to differential chromatin are already known to play an important role within the molecular clock, and are recognized to be important for entrainment and SD. In addition several novel genes with differential chromatin nearby have been identified that also likely have an important role during entrainment and following SD. Several areas of differential chromatin were investigated using a luciferase reporter assay, and results obtained indicate the areas of chromatin are able to strongly drive gene expression, and respond to treatment using forskolin and horse serum, both known to have an effect on the molecular clock.

Proteins are integral to all cellular processes and underpin the function of all extant organisms, meaning variants impacting them are a primary cause of phenotypic variation. Protein coding variants are a key area of study in biology, with relevance from structural and molecular biology to population genetics. They are also medically important, impacting inherited genetic diseases, cancer and response to pathogens. Recent advances in highthroughput experimental techniques have opened the door to many new approaches in biology, and protein variants are no exception. Deep mutational scanning experiments exhaustively measure the fitness of variants in a protein, which gives us more experimentally validated mutational consequence measurements than ever before. Such advances, together with ever larger sequence and structure databases, have created an opportunity to apply large scale analyses to coding variation, studying the effect on protein structure, function and phenotype. In this thesis I perform three large scale variant analyses. First, I use the consequences of variation to learn about protein structure and function. I compile a dataset from 28 deep mutational scanning studies, covering 6291 positions in 30 proteins, and use the consequences of mutation at each position to define a mutational landscape. I show rich biophysical relationships in this landscape and identify functionally distinct positional subtypes of each amino acid. In the second analysis, I explore genotype to phenotype prediction using a dataset of 1011 S. cerevisiae strains, with genotypes, transcriptomics, proteomics and measured phenotypes, and comprehensive gene deletions in four strains. I show knowledge-based models of mutational consequences and pathway function can be used to associate genes with phenotypes and predict growth phenotypes across 34 growth conditions. However, genetic background is found to have a large effect on variant consequences, to such an extent that the same deletion can be highly significant in one strain and have no effect in another. Finally, I analyse computational variant effect prediction, benchmarking current predictors using deep mutational scanning data. I then develop a new end-to-end deep convolutional neural network predictor that predicts consequences directly from sequence and structure and show it improves on current methods. Together these projects advance our knowledge of protein coding variation and enhance our capacity to link variation to impacts on structure, function and phenotype.

A bacterial species' genetic content can be remarkably fluid. The collection of genes found within a given species is called the pan-genome and is generally much larger than the gene repertoire of a single cell. A consequence of this pan-genome is that bacterial genomes are highly adaptable and thus variable. The dominant paradigm for analysing genetic variation relies on a central idea: all genomes in a species can be described as minor differences from a single reference genome, which serves as a coordinate system. As an introduction to this thesis, we outline why this approach is inadequate for bacteria and describe a new approach using genome graphs. In the first chapter, we present algorithms for de novo variant discovery within such genome graphs and evaluate their performance with empirical data. The remaining chapters address a question relating to a critical bacterial pathogen: can Nanopore sequencing of Mycobacterium tuberculosis provide high-quality public health information? We collect data from Madagascar, South Africa, and England to help answer this question. First, we assess outbreaks identified using single-reference and genome graph methods. Second, we evaluate antimicrobial resistance predictions and introduce a framework for using genome graphs to improve current methods. Lastly, we train an M. tuberculosis-specific Nanopore basecalling model with considerable accuracy improvement. Together, this thesis provides general methods for uncovering bacterial variation and applies them to an important global public health question.