Oxford Nanopore Technologies (ONT) sequencing offers real-time, long-read capabilities for pathogen genomics, yet concerns persist regarding its accuracy in critical applications like outbreak tracking. This dissertation evaluates ONT-based workflows for genome reconstruction, high-resolution genotyping, and molecular characterization, presenting a scalable, reproducible bioinformatics pipeline for bacterial whole-genome sequencing (WGS). Genome reconstruction and ONT specific error patterns were investigated across a diverse panel of zoonotic and healthcare-associated pathogens, including Bacillus anthracis, Brucella spp., and Klebsiella spp. Results show that assembly with Flye followed by Medaka polishing (using bacteria-specific models) provides the highest accuracy. However, persistent biases in homopolymer-rich, high-GC, and methylated regions highlight a continued need for algorithmic improvements. Regarding genotyping, both cgMLST and SNP-based approaches were assessed. Despite ONT's higher error rate, optimized long-read workflows reliably reconstructed outbreak clusters. Furthermore, in silico serotyping of Salmonella enterica with nanopore sequencing data showed high concordance with classical assays. The framework’s utility was demonstrated via a case study of Streptococcus uberis outbreaks in Thuringia. The analysis identified a resistance-carrying plasmid linked to bovine mastitis, a feature only resolvable through ONT sequencing. Despite these advantages, ONT’s evolving hardware, chemistry, and neural network-based basecallers introduce reproducibility challenges, particularly for rare taxa underrepresented in training datasets. This work underscores ONT’s strong potential as a standalone platform for rapid, field-deployable analysis of pathogen genomics, while emphasizing the need for continued benchmarking and careful consideration of sequencing biases in epidemiological applications.
Christine Thomas (Thu,) studied this question.