Highly pathogenic avian influenza H5Nx clade 2.3.4.4b was introduced to North America in 2021, causing widespread illness in domesticated and wild animals. Although the current burden of human disease is low (74 human cases across Canada, the United States and Mexico), avian influenza has a high historical case fatality rate. Additionally, clade 2.3.4.4b represents a critical pandemic concern due to its unprecedented mortality in wild birds and expanded mammalian host range. This study will explore the genomic basis of this range by identifying mutations significantly associated with sampled hosts. These putative host-determining mutations will inform risk assessment of emerging H5Nx lineages and direct the functional characterization of key mutations. 18,460 complete H5Nx genomes were extracted and cleaned from public repositories, NCBI and GISAID. Given the presence of 253 H5Nx host labels, we first sought to empirically combine species into host groups sharing viral adaptations. Hosts were combined based on their co-occurrence in InfoMap and Spectral clusters based on viral genetic distances. Unfortunately, this yielded groupings not supported by the literature (e.g., foxes with ducks) suggesting host group determination likely would benefit from incorporating host evolution and ecology. Using host taxonomic orders instead, we then performed genome-wide association studies using pyseer and treeWAS across all eight genome segments. Preliminary results indicate the presence of mutations significantly associated with the majority of host orders. Future work will use PRSice to calculate the polygenic risk score using significant mutations across segments to identify viruses with the greatest risk of infecting hosts’ taxonomic order.

Sin Nombre Virus (SNV), a rodent-borne zoonotic hantavirus, causes Hantavirus Cardiopulmonary Syndrome (HCPS) in North America. Deer mice (Peromyscus maniculatus), the primary reservoir, inhabit diverse environments, including peridomestic and agricultural habitats where most HCPS cases are acquired. While SNV prevalence and HCPS outbreaks can sometimes be predicted by climate and vegetation changes in natural systems, these relationships are poorly understood in agricultural settings. Agricultural disturbances, such as harvesting and grazing, alter vegetation structure, impacting small mammal community composition, population dynamics, and movement, and potentially altering SNV transmission risk. We conducted a multi-season capture-mark-recapture study across crop fields, hay fields, and cattle pastures in Alberta, Canada to assess how land-use type and disturbance regime influence small mammal community composition, host abundance, and SNV prevalence. Deer mice and meadow voles (Microtus pennsylvanicus) dominated communities, but relative abundance varied by field type. Crop fields were almost exclusively inhabited by deer mice, whereas meadow voles dominated pastures. Across all field types, deer mouse abundance consistently peaked with vegetation cover. In 2024, SNV was only detected in deer mice inhabiting crop fields late in the growing season. Highest SNV prevalence followed both peak deer mouse abundance and harvesting disturbance. These findings identify post-harvest crop fields in early fall as high-risk settings for SNV occurrence. By linking agricultural disturbance, host ecology, and pathogen surveillance, our research demonstrates a One Health approach to zoonotic risk assessment. Our findings demonstrate how specific farming practices shape disease risk, providing insights for targeted SNV surveillance and mitigation in agricultural landscapes.

Antimicrobial resistance (AMR) in emerging pathogens such as Candida auris (C. auris) poses a major global health threat, yet treatment failure is often driven not only by genetic resistance but also by transient antifungal tolerance. Here, we develop a deterministic population-level model to investigate the coupled dynamics of susceptible, tolerant, and resistant C. auris subpopulations under antifungal treatment and innate immune pressure. The model incorporates drug class (static vs. cidal), infection load, and immune activity to quantify their combined effects on population dynamics, resistance emergence, and infection clearance. Our results show that antifungal tolerance dominates early treatment dynamics across a wide range of conditions, frequently acting as the primary surviving phenotype and a reservoir for subsequent resistance evolution. High infection loads delay clearance and promote phenotypic diversification, while stronger innate immunity accelerates population collapse and suppresses both tolerance and resistance. We further identify drug-dependent differences: static drugs promote rapid tolerant expansion followed by resistance, whereas cidal drugs delay resistance emergence but sustain prolonged tolerance. Importantly, increasing drug concentration yields diminishing returns beyond a threshold, with limited impact on long-term evolutionary outcomes. These findings highlight that treatment efficacy depends critically on the interplay between drug action, infection burden, and host immunity. Our work suggests that strategies enhancing immune–drug synergy may be more effective than dose escalation alone. This framework provides quantitative insight into antifungal treatment failure and offers guidance for optimizing therapeutic interventions against multidrug-resistant fungal infections.

Many researchers and trainees wonder how their skills and experience would translate to careers in the private sector. Drawing on my experience transitioning from academia to a senior software engineering role, this presentation will explore the opportunities available to individuals with research backgrounds and the skills that contribute to success in both academic and industry environments. I will discuss examples of research and development work in corporate settings, compare how expectations, incentives, and measures of success differ between academia and industry, and highlight the often under-valued “soft skills” of communication, problem-solving, and collaboration that are developed through research training. Finally, I will examine how generative AI is reshaping day-to-day work in software engineering and consider what these changes may mean for researchers and computational scientists pursuing careers in technology.

CDBProm v2.0 represents a large-scale database containing millions of AI-predicted bacterial promoters with functional annotation. The predictive framework used to build CDBProm v2.0 consists of a fine-tuned DNABERT6 model, which uses the same logics of Large Language Models such as ChatGPT (OpenAI), Gemini (Google), etc. The resulting model was accurate in modelling the long-range dependencies of transcription factor binding sites that govern bacterial transcription. CDBProm v2.0 is free to use and fully open access (https://cdbprom2-iimas.unam.mx), and provides a modern framework for understanding bacterial transcription regulation and annotating genomes.

In prokaryotes lateral gene transfer (LGT) drives intraspecific variation in gene content and the formation of pangenomes. In microbial eukaryotes, however, the extent to which LGT contributes to pangenome structure remains unclear. Given the diversity of viruses that infect protists, viruses have been hypothesized to mediate LGT in eukaryotes, and evidence for virus-mediated gene exchange has been found in several protist species. To investigate the interplay between LGT, pangenome evolution and viral interactions in microbial eukaryotes, we generated high-quality, long-read sequencing based genome assemblies for stramenopiles; this includes pelagophytes, including multiple strains of the harmful algal bloom-forming species Aureococcus anophagefferens, and diverse thraustochytrids – a phylum recently identified as a natural host for newly discovered mirusviruses and other giant viruses. Comparative genomics and phylogenetics reveled remarkable strain level genetic variation in A. anophagefferens, with a pangenome comprised of 23,356 orthogroups (81.1% core, 18.9% accessory). Although gene content variation does not appear to be primarily driven by recent prokaryotic LGTs (2.6% of accessory orthogroups), we identified 368 orthogroups of bacterial origin that were acquired in a common ancestor of all analyzed strains and were not found in other pelagophyte algae, suggesting a role in the ecological success and bloom persistence of Ac. anophagefferens. We also find some evidence for viral involvement shaping strain-level genomic variation in pelagophytes. Finally, we establish an initial comparative framework for thraustochytrids genome evolution and their virus-host interactions, including the existence of a viral-pangenome within this system.

Inferring species presence from the genetic material they leave behind in the environment has revolutionized biodiversity monitoring. This technique, known as environmental (e)DNA analysis, offers a non-invasive alternative to traditional methods and can be applied to a wide range of taxa. The use of universal primers (“metabarcoding”) allows multiple species to be detected from a single analysis, thereby offering avenues to assess community composition. One of the largest benefits of eDNA analyses, however, can also be one of its largest challenges: a lack of visual confirmation. This challenge is compounded by the nature of eDNA, which often exists in low concentrations and is susceptible to environmental degradation. Determining ecological realities from metabarcoding data requires careful interpretation of obtained eDNA sequences, particularly when a single amplicon sequence variant (ASV) matches to multiple plausible species. In this talk, I will discuss the specific bioinformatic thresholds and manual checks I employ in my research examining fish community composition in coastal ocean and freshwater ecosystems.

Although microorganisms are central to all ecosystems, their symbiotic relationships and ecological roles remain poorly understood. This work explores microbial interactions and how they shape ecosystems and offer biotechnological potential. Using a multidisciplinary and multi-omics approach, we explored the composition and function of the bacterial microbiome associated with O. marina. Microbiome and metagenomic analyses were employed to identify microbiome composition, while classical microbiology combined with whole-genome sequencing enabled in-depth characterization of culturable microorganisms. Interactions between host and microbiome were examined through chemotaxis, co-culture experiments, metabolic network predictions, and RNA-seq–based differential gene expression analyses. Results revealed that the microbiome of O. marina comprises more than 100 bacterial species. Behavioral assays revealed that O. marina displayed both attraction and avoidance responses depending on the bacterial partner. Growth in co-culture also varied, with some bacteria negatively affecting host performance, suggesting that these microbiomes rely on a finely balanced composition rather than simple mutualism. Metabolic predictions and gene expression analyses point to complex interactions driven by complementary metabolic pathways between organisms. Additionally, genomic analysis of culturable bacteria uncovered a wide range of potential applications, including the production of antifungal compounds, degradation of hydrocarbons and phthalates, reduction of heavy metals, biomineralization processes, and the synthesis of essential cofactors and vitamins such as cobalamin. This work suggests that microbiomes associated with microeukaryotes are far more complex than previously recognized. They represent rich reservoirs of microbial diversity with significant biotechnological potential and may play essential roles in the modulation, transformation, and resilience of marine ecosystems.

Pathogen diagnostics are fundamental to clinical decision-making, enabling targeted therapeutic interventions through accurate identification of infectious agents and assessment of antimicrobial resistance (AMR). Traditionally, the diagnostics pipeline required 48-72 hours, due to reliance on culture-based techniques that involve multiple overnight incubation steps, which delays treatment in critical cases. Here, we leverage a plasmonically enhanced microfluidic platform integrating loop-mediated isothermal amplification (LAMP), broth microdilution (BMD), and nanostructure-enabled sensing for rapid and multiplexed bacterial identification and phenotypic AMR screening. The Plasmonic elements accelerate the LAMP and BMD assay kinetics via novel optical phenomena, including plasmonic catalysis and structural colorimetry, achieving a sample-to-answer time of 36 minutes. The system identification performance was validated with LAMP assays targeting Escherichia coli, Enterococcus spp., Pseudomonas aeruginosa, Staphylococcus aureus, Streptococcus agalactiae, and Streptococcus pneumoniae. In parallel, on-chip BMD assays were evaluated against a panel of 12 antibiotics using both Gram-positive and Gram-negative strains. The platform demonstrated 100% identification concordance and 100% essential agreement in AMR profiling relative to gold-standard MALDI-TOF and conventional BMD assays. Overall, this work establishes a rapid, accurate, and scalable diagnostic platform that significantly reduces turnaround time while maintaining gold-standard performance.

Background: In livestock farming, variability in the implementation of biosecurity practices influences disease prevention and control. Clustering techniques can identify groups of farms with similar implementation patterns. Our objective was to identify biosecurity implementation profiles among Canadian dairy farms using ensemble clustering. Material and Methods: A cross-sectional study was conducted on 346 Canadian dairy farms between 2022 and 2023. Biosecurity practices were assessed using the Risk Assessment Questionnaire (RAQ), conducted as part of the proAction program, which evaluates the implementation of biosecurity practices using four categorical levels (never, sometimes, often, and always). Additional farm demographic and management characteristics were collected during study visits. A total of 296 farms with available RAQ data were included (176 Québec, 70 Ontario, and 50 Alberta). Multiple clustering solutions derived from dimensionality reduction and clustering algorithms were integrated within an ensemble framework using consensus clustering. Results: Ensemble clustering identified three biosecurity implementation profiles. Profile A (n=146) showed frequent implementation of multiple practices and included farms from all three provinces. Profile B (n=71) showed selective implementation, with some practices strongly represented while others—particularly disease-management practices—were frequently absent, and included only farms from Ontario and Alberta. Profile C (n=79) showed limited implementation and was composed almost exclusively of Québec farms. Housing type and milking system differed across profiles. Conclusion: Identifying biosecurity implementation profiles provides a structured way to interpret RAQ results and highlights heterogeneity in on-farm practices. These profiles may support risk-based animal movement decisions by complementing individual animal testing with farm-level biosecurity data.

Tuberculosis (TB), caused by mycobacterium tuberculosis (Mtb), is the number one infectious disease killer and is the most common cause of death in people living with HIV (PLWH). HIV and TB co-infection place an immense burden on health care systems as they act in synergy to worsen each other’s disease prognosis. Another factor complicating TB care is the increasing prevalence of multi- and extensively-drug resistant TB (MDR/XDR). While HIV and TB are endemic in sub-Saharan Africa, they also disproportionately affect marginalized populations in Canada. Unfortunately, the only licensed TB vaccine, BCG, doesn’t protect from adult pulmonary TB and is not recommended for PLWH. The best way to prevent the emergence of MDR/XDR TB is to develop novel efficacious TB vaccines, which are safe and effective to use in those who are highly at risk. Next-generation humanized mice are ideal models to research this as they recapitulate a more complete human immune response and can be successfully infected with HIV, TB and HIV/TB. We have found that they recapitulate many aspects of human HIV and TB disease pathology. We investigated the immunogenicity and protection efficacy of a next-generation respiratory mucosal trivalent chimpanzee adenoviral-vectored vaccine (Tri:ChAd:TB) in naïve and HIV-infected humanized mice. When immunizing naïve mice, a trend of increased Mtb-specific CD4+ T cells producing IFNy and TNFα in the lungs and spleen was observed. Vaccinated humanized mice challenged with Mtb exhibited significantly reduced lung mycobacterial burden, decreased tissue dissemination and improved lung pathology. HIV-infected mice that were subsequently immunized and challenged with Mtb had a decreased trend in mycobacterial load in the lungs compared to unvaccinated, indicating that the vaccine may offer protection against TB, even in the context of HIV infection. These findings demonstrate the efficacy of the respiratory mucosal Tri:ChAd:TB vaccine and could halt this global epidemic by preventing further emergence of MDR/XDR TB.

Invasive group A streptococci (iGAS) infections cause significant morbidity and mortality worldwide. Following the COVID-19 pandemic, iGAS activity increased in Nova Scotia and associations were made with the rise of a hypervirulent strain called M1UK. However, M1UK only partially explained the increased iGAS activity. As macrolide resistance also increased during the post-pandemic period, our laboratory used phenotypic and genetic methods to investigate whether macrolide resistance could also have contributed to the increased iGAS activity. As you will see in the presentation, there was a concomitant rise of hypervirulent M1UK and macrolide-resistant strains (emm58, emm77, emm83, and emm92) in recent years. In contrast to national rate, macrolide resistance in Nova Scotia has expanded rapidly since 2023 and was largely attributed to emm92 containing a plasmid-borne inducible ermT gene. The proportions of emm92 were much higher than seen nationally, suggesting local clonal expansion and as macrolide and tetracycline resistance genes are both found in iGAS isolates, co-selection of antibiotic resistance determinants is possible. Epidemiologic studies like this allows us to speculate possible causes of drivers of antibiotic resistances, and help shape further investigations.

Each year, approximately three million newborns die from sepsis worldwide, with ~75% of all deaths under-five years occurring within the first week of life. Sepsis is a global concern, especially to young infants who are at highest risk of mortality. However, there lacks a quick and reliable way to identify the infectious microbes, resulting in babies being over- or under-treated with antibiotics that may contribute to resistance and depletion of limited resources. To tackle this challenge, our group is applying systems biology approaches and advanced omics analyses within a Global Health context to: (1) define host molecular markers in bacterial sepsis using genome-wide transcriptomic analyses and machine learning; (2) identify potential virulence factors in the causal bacterial pathogens using microbial genome-wide association studies; and (3) apply cutting-edge long-read Nanopore 16S sequencing to develop vaccines and improve pathogen identification. Ultimately, our goal is to understand why newborns are highly susceptible to infections during their first week of life, and apply advanced omics strategies to develop real-world applications including vaccines, diagnostics and therapeutics.

This presentation will outline major antimicrobial drug ranking, prioritization and categorization systems and lists, with a focus on their goals, gaps and how they can be used to support stewardship and surveillance activities.