Computational Surveillance and Molecular Analysis of Lassa Virus in Response to Increased Prevalence in Humans and Animals

Lassa fever, first identified in the late 1960s, is caused by Lassa virus (LASV), a Biosafety Level 4 (BSL-4) pathogen endemic to West Africa. It is estimated to cause approximately 300,000 infections and 5,000 deaths annually. Despite its public health impact, no licensed vaccine is currently available. Exported cases have been reported across multiple continents, including Europe and China, and increasing global mobility and environmental change may facilitate the establishment of LASV in new regions. While the multimammate rat (Mastomys natalensis) is the primary reservoir, recent reports indicate potential circulation in other animal hosts, potentially increasing human infections. This emphasises the urgent need for improved surveillance and better understanding of the virus—the central aim of this thesis. Distinct LASV lineages circulate in specific geographic regions, show varying immunological behaviors, and may contribute to differing disease outcomes. Hence, tracking of the lineages offers actionable insight for epidemic control. This thesis presents CLASV, a command-line tool based on random forest (RF) classification for rapid LASV lineage assignment from nucleotide sequences. The tool incorporates a customised out-of-distribution detection framework to distinguish LASV from other mammarenaviruses. In parallel, a suite of open-access genomic surveillance resources was developed using the Nextstrain and Nextclade platforms, enabling real-time phylogenetic visualisation and mutation tracking. Together, these tools provide a complementary framework for integrating LASV genomics into outbreak response and public health surveillance. Curated data derived from CLASV and Nextstrain enabled a population-scale molecular analysis investigating biophysical differences among LASV lineages. These descriptors include fundamental properties such as protein length, molecular mass, and amino acid composition, which define the physical state of viral proteins in nature. By employing an integrative approach combining RF-feature importance, distance profiling, correlation analysis, and mass-weighted amino acid composition, lineage-specific amino acid preferences were identified. Establishing the distributions of these properties provides a baseline for interpreting historical variation and enables the detection of future deviations, including the emergence of novel variants or atypical protein features that may warrant functional or public health attention. In studying these differences, a comprehensive catalogue of predicted LASV glycoprotein structures was generated and publicly deposited to support ongoing structural and immunological studies. Finally, this work highlights a global imbalance in high-containment research capacity. Although Africa harbors a high diversity of endemic BSL-4 pathogens, it possesses only a limited proportion of global BSL-4 infrastructure. As a likely consequence of this, LASV genomic data are generated outside endemic regions, with delays often exceeding two years between sample collection and public deposition. Despite over four decades of recurrent outbreaks, far fewer than 4,000 high-quality LASV genomes are currently available. This scarcity constrains genomic analyses and represents a critical limitation for epidemic preparedness and global health security.