Fungal infections of animals have become an increasingly important global issue. White-Nose Syndrome is an ongoing fungal epizootic of North American hibernating bats, caused by epidermal infections of the fungus, Pseudogymnoascus destructans. Infections emerged early in 2006 in New York State and have since spread to 35 US States and seven Canadian Provinces, with rates of mortality exceeding 90% in some bat colonies. As an emerging outbreak in North America, the transmission of P. destructans is assumed to occur in a radial fashion outwards from the point of origin. In addition, the factors that may influence P. destructans transmission have been postulated, but not tested before. Lastly, as reproduction is assumed to be strictly clonal in North America, invasive populations should have low genetic diversity, and may even accumulate deleterious mutations over time. The aim of my PhD research is to test these assumptions regarding the spread, evolution, and adaptation of P. destructans using combination of genotyping methods. My results showed how P. destructans isolates have shifted in terms of phenotypes and physiological capabilities since being introduced. In addition, I describe patterns of connectivity across the landscape, which are more consist with the level of anthropogenic activity than variation in climate. The mutations common to all invasive strains of P. destructans are associated with adaptations that have occurred since being introduced from Europe, some with relevant metabolic functions that fit their pathogenic lifestyle. Together, my results revealed significant phenotypic and genotypic changes during the spread of P. destructans in North America. The factors identified here that influence the phenotypic and genotypic changes should help developing better management strategies against the White-Nose Syndrome pathogen. / Thesis / Doctor of Philosophy (PhD)

This research investigates the application of machine learning techniques in computational genomics across two distinct domains: (1) the predicting the source of bacterial pathogen using whole genome sequencing data, and (2) the functional annotation of genes using single- cell RNA sequencing data. This work proposes the development of a bioinformatics pipeline tailored for identifying genomic variants, including gene presence/absence and single nu- cleotide polymorphism. This methodology is applied to specific strains such as Salmonella enterica serovar Typhimurium and the Ralstonia solanacearum species complex. Phylo- genetic analyses along with pan-genome and positive selection studiesshow that genomic variants and evolutionary patterns of S. Typhimurium vary across sources, which suggests that sources can be accurately attributed based on genomic variants empowered by machine learning. We benchmarked seven traditional machine learning algorithms, achieving a no- table accuracy of 94.6% in host prediction for S. Typhimurium using the Random Forest model, underscored by SHAP value analyses which elucidated key predictive features. Next, the focus is shifted to the prediction of Gene Ontology terms for Arabidopsis genes using single-cell RNA-seq data. This analysis offers a detailed comparison of gene expression in root versus shoot tissues, juxtaposed with insights from bulk RNA-seq data. The integration of regulatory network data from DAP-seq significantly enhances the prediction accuracy of gene functions. / Master of Science / This work applies machine learning techniques to two areas in computational biology: pre- dicting the hosts of bacterial pathogens based on their genome data, and predicting the func- tions of plant genes using single-cell gene expression data. The first part develops a method to analyze genome sequences from bacterial pathogens like Salmonella enterica serovar Ty- phimurium and the Ralstonia solanacearum species complex, identifying genomic variants, including gene presence/absence and single nucleotide polymorphism, which are variations in genetic code. By studying the evolutionary relationships and genetic diversity among dif- ferent strains, the motivation for using machine learning models to predict the sources (e.g., poultry, swine) of the pathogen genomes is established. Several machine learning models are then trained on these datasets, and the most important factors contributing to the predic- tions are identified. The second part focuses on predicting the functions of genes in the model plant species Arabidopsis thaliana using the gene expression data measured at the single-cell level to train machine learning models for identifying standardized gene function descrip- tions called Gene Ontology (GO) terms. By comparing results from single-cell and bulk tissue data, the study evaluates whether the higher resolution of single-cell data improves gene function prediction accuracy. Additionally, by incorporating information about gene regulation from a specialized experiment, the role of gene expression control in determining gene functions is explored.

Yes / Aeromonas hydrophila has emerged worldwide as a human pathogen. Here, we report the draft whole-genome sequence of a freshwater isolate from Malaysia, A. hydrophila strain M062, and its N-acylhomoserine lactone genes are also reported here. / University of Malaya via High-Impact Research Grants (UM C/625/1/HIR/MOHE/CHAN/01, no. A-000001- 50001), and aUM-MOHEHIR grant (UM C/625/1/HIR/MOHE/CHAN/ 14/1, no. H-50001-A000027)

Yes / Trypanosomes cause a variety of diseases in man and domestic animals in Africa, Latin America, and Asia. In the Trypanozoon subgenus, Trypanosoma brucei gambiense and Trypanosoma brucei rhodesiense cause human African trypanosomiasis, whereas Trypanosoma brucei brucei, Trypanosoma evansi, and Trypanosoma equiperdum are responsible for nagana, surra, and dourine in domestic animals, respectively. The genetic relationships between T. evansi and T. equiperdum and other Trypanozoon species remain unclear because the majority of phylogenetic analyses has been based on only a few genes. In this study, we have conducted a phylogenetic analysis based on genome-wide SNP analysis comprising 56 genomes from the Trypanozoon subgenus. Our data reveal that T. equiperdum has emerged at least once in Eastern Africa and T. evansi at two independent occasions in Western Africa. The genomes within the T. equiperdum and T. evansi monophyletic clusters show extremely little variation, probably due to the clonal spread linked to the independence from tsetse flies for their transmission. / Funding was received from the Research Foundation Flanders (FWO, grants 1501413N and 1101614N) and the European DG Health and Food Safety (SANTE). We thank the Center of Medical Genetics at the University of Antwerp for hosting the NGS facility.

Yes / Background: Tracking recent transmission is a vital part of controlling widespread pathogens such as Mycobacterium tuberculosis. Multiple methods with specific performance characteristics exist for detecting recent transmission chains, usually by clustering strains based on genotype similarities. With such a large variety of methods available, informed selection of an appropriate approach for determining transmissions within a given setting/time period is difficult. Methods: This study combines whole genome sequence (WGS) data derived from 324 isolates collected 2005–2010 in Kinshasa, Democratic Republic of Congo (DRC), a high endemic setting, with phylodynamics to unveil the timing of transmission events posited by a variety of standard genotyping methods. Clustering data based on Spoligotyping, 24-loci MIRU-VNTR typing, WGS based SNP (Single Nucleotide Polymorphism) and core genome multi locus sequence typing (cgMLST) typing were evaluated. Findings: Our results suggest that clusters based on Spoligotyping could encompass transmission events that occurred almost 200 years prior to sampling while 24-loci-MIRU-VNTR often represented three decades of transmission. Instead, WGS based genotyping applying low SNP or cgMLST allele thresholds allows for determination of recent transmission events, e.g. in timespans of up to 10 years for a 5 SNP/allele cut-off. Interpretation: With the rapid uptake of WGS methods in surveillance and outbreak tracking, the findings obtained in this study can guide the selection of appropriate clustering methods for uncovering relevant transmission chains within a given time-period. For high resolution cluster analyses, WGS-SNP and cgMLST based analyses have similar clustering/timing characteristics even for data obtained from a high incidence setting. / ERC grant [INTERRUPTB; no. 311725] to BdJ, FG and CJM; an ERC grant to TS [PhyPD; no. 335529]; an FWO PhD fellowship to PM [grant number 1141217N]; the Leibniz Science Campus EvolLUNG for MM and SN; the German Centre for Infection Research (DZIF) for TAK, MM, CU, PB and SN; a SNF SystemsX grant (TBX) to JP and TS and a Marie Heim-Vögtlin fellowship granted to DK by the Swiss National Science Foundation. The computational resources and services used in this work were provided by the VSC (Flemish Supercomputer Center), funded by the Research Foundation - Flanders (FWO) and the Flemish Government – department EWI.

Yes / We investigated whether companion drug resistance was associated with adverse outcome of the shorter MDR-TB regimen in Bangladesh, after adjusting for fluoroquinolone resistance. MDR/RR-TB patients registered for treatment with a standardized gatifloxacin-based shorter MDR-TB regimen were selected for the study. Drug resistance was determined using the proportion method, gatifloxacin and isoniazid minimum inhibitory concentration testing for selected isolates, and whole genome sequencing. Low-level and high-level fluoroquinolone resistance were the most important predictors of adverse outcomes, with pyrazinamide resistance having a significant yet lower impact. In patients with fluoroquinolone-/second-line injectable-susceptible TB, non-eligibility to the shorter MDR-TB regimen (initial resistance to either pyrazinamide, ethionamide, or ethambutol) was not associated with adverse outcome (aOR 1.01; 95%CI 0.4-2.8). Kanamycin resistance was uncommon (1.3%). Increasing levels of resistance to isoniazid predicted treatment failure, also in a subgroup of patients with high-level fluoroquinolone-resistant TB. Our results suggest that resistance to companion drugs of the shorter MDR-TB regimen, except kanamycin resistance, is of no clinical importance as long as fluoroquinolone susceptibility is preserved. Hence, contrary to current WHO guidelines, exclusions to the standard regimen are justified only in the case of fluoroquinolone, and possibly kanamycin resistance. / Damien Foundation Belgium for its financial and logistic support to run the project including its research activities. European Research Council (Starting Grant INTERRUPTB 311725).

Yes / BACKGROUND: Whole-genome sequencing (WGS) of Mycobacterium tuberculosis allows rapid, accurate inferences about the sources, location and timing of transmission. However, in an era of heightened concern for personal privacy and science distrust, such inferences could result in unintended harm and undermine the public´s trust. METHODS: We held interdisciplinary stakeholder discussions and performed ethical analyses of real-world illustrative cases to identify principles that optimise benefit and mitigate harm of M. tuberculosis WGS-driven TB source investigations.RESULTS: The speed and precision with which real-time WGS can be used to associate M. tuberculosis strains with sensitive information has raised important concerns. While detailed understanding of transmission events could mitigate harm to vulnerable patients and communities when otherwise unfairly blamed for TB outbreaks, the precision of WGS can also identify transmission events resulting in social blame, fear, discrimination, individual or location stigma, and the use of defaming language by the public, politicians and scientists. Public health programmes should balance the need to safeguard privacy with public health goals, transparency and individual rights, including the right to know who infects whom or where.CONCLUSIONS: Ethical challenges raised by real-time WGS-driven TB source investigation requires public health authorities to move beyond their current legal mandate and embrace transparency, privacy and community engagement.

Infectious diseases cause substantial morbidity and mortality, exacerbated by increasing antibiotic resistance. In critically ill patients, recent studies indicate a substantial variability in β-lactam antibiotic levels when standardized dosing is applied. New methods for characterizing nosocomial outbreaks of bacterial infections are needed to limit transmission. The goals of this thesis were to investigate new strategies towards individualized treatment and control of bacterial infections.  In Paper I we confirmed high variability in β-lactam antibiotic levels among intensive care unit (ICU) patients from southeastern Sweden, where 45 % failed to reach treatment targets (100 % fT>MIC). Augmented renal clearance and establishing the minimum inhibitory concentration of the bacteria were important for evaluating the risk of not attaining adequate drug levels. In Paper II a rapid ultra-performance liquid chromatography tandem mass spectrometry (UPLC-MS/MS) method for simultaneous quantification of 11 commonly used antibiotics was developed and tested in clinical samples. Performance goals (CV<15%) were reached. A microbiological method for quantification of β-lactam antibiotics in serum was developed in Paper III. The method could be important for hospitals without access to an LC-MS method. Paper IV and Paper V investigated ligation-mediated qPCR with high resolution melt analysis (LMqPCR HRMA), for transmission investigation of extended spectrum β-lactamase (ESBL)-producing E. coli and other common bacterial pathogens. Results comparable to the reference method (PFGE) could be achieved within one day in a closed system and confirmed a nosocomial outbreak in Kalmar County. In Paper VI whole genome sequencing followed by bioinformatic analysis resolved transmission links within a nosocomial outbreak due to improved discriminatory power compared to LMqPCR HRMA. The high proportion of ICU patients with insufficient β-lactam drug levels emphasizes the need for individualized treatment by therapeutic drug monitoring (TDM). TDM is enabled by a highly sensitive method, such as UPLC-MS/MS, but if unavailable, also by a microbial method. Molecular typing methods used for transmission investigation can detect nosocomial outbreaks. LMqPCR HRMA can be used for screening purposes. For enhanced resolution, whole genome sequencing should be used, but always together with a rigorous epidemiological investigation.

Next-generation sequencing methods now allow rapid and cost-effective sequencing of DNA on a scale not previously possible. This offers great opportunities for the research of Mendelian disorders, but also significant challenges. The sequencing of exomes, or whole genomes, has emerged as a powerful clinical research tool, with targeted gene analyses generally being preferred in the clinical diagnostic setting. These methods have been employed here with the aim of identifying novel genetic causes of inherited heart disorders and to gain insights into the utility and limitations of these techniques for clinical diagnosis in these disorders. Data produced from the introduction of a targeted multi-gene next-generation sequencing test into clinical practice has been studied. Variation within the mitochondrial genome has been analysed to assess the importance of mitochondrial DNA variants in patients with hypertrophic cardiomyopathy. The m.4300A>G mutation is identified as an important cause of this disorder, with other previously cardiomyopathy-associated and novel variants also identified. Such multi-gene tests can facilitate interpretable and phenotype-relevant results, but at the expense of limiting more extensive data acquisition. Whole-genome sequencing has been performed in five families with different autosomal dominant inherited heart disease phenotypes of unknown genetic aetiology. In two of these likely pathogenic variants were identified, one in the gene encoding titin (TTN) and the other in the calcium channel subunit gene CACNA1C. In vitro studies were undertaken to support the pathogenicity of the TTN variant and understand the functional effects of this. In the other three families either multiple candidate gene variants were identified or no clear candidate variant was identified. This highlights the difficulties in interpreting these results, even in carefully selected families. Overall, although the research benefits of exome or genome studies are evident, the interpretation and validation of genetic variant data produced remains highly challenging for clinical diagnosis.

In recent years, the genetic landscape of B-cell derived lymphoid malignancies, including chronic lymphocytic leukemia (CLL), has been rapidly unraveled, identifying recurrent genetic mutations with potential clinical impact. Interestingly, ~30% of all CLL patients can be assigned to more homogeneous subsets based on the expression of a similar or “stereotyped” B-cell receptor (BcR). Considering that biased distribution of genetic mutations was recently indicated in specific stereotyped subsets, in paper I, we screened 565 subset cases, preferentially assigned to clinically aggressive subsets, and confirm the SF3B1 mutational bias in subset #2 (45%), but also report on similarly marked enrichment in subset #3 (46%). In contrast, NOTCH1 mutations were predominantly detected in subsets #1, #8, #59 and #99 (22-34%). This data further highlights a subset-biased acquisition of genetic mutations in the pathogenesis of at least certain subsets. Aberrant NF-κB signaling due to a deletion within the NFKBIE gene previously reported in CLL warranted extended investigation in other lymphoid malignancies. Therefore, in paper II, we screened 1460 patients with various lymphoid malignancies for NFKBIE deletions and reported enrichment in classical Hodgkin lymphoma (27%) and primary mediastinal B-cell lymphoma (PMBL) (23%). NFKBIE-deleted PMBL cases had higher rates of chemorefractoriness and inferior overall survival (OS). NFKBIE-deletion status remained an independent prognostic marker in multivariate analysis. EGR2 mutations were recently reported in advanced stage CLL patients; thus, in paper III we screened 2403 CLL patients for mutations in EGR2. An overall mutational frequency of 3.8% was reported and EGR2 mutations were associated with younger age, advanced stage and del(11q). EGR2 mutational status remained an independent marker of poor outcome in multivariate analysis, both in the screening and validation cohorts. Whole-genome sequencing (WGS) of 70 CLL cases, assigned to poor-prognostic subsets #1 and #2 and indolent subset #4, were investigated in Paper IV and revealed a similar skewing of SF3B1 mutations in subset #2 and NOTCH1 mutations in subset #1 to that reported in Paper I. Additionally, an increased frequency of the recently proposed CLL driver gene RPS15 was observed in subset #1. Finally, novel non-coding mutational biases were detected in both subset #1 and #2 that warrant further investigation.