Oxford Nanopore User Group Meeting, Hong Kong

Accurate and timely prediction of antimicrobial resistance (AMR) is crucial for guiding optimal antibiotic selection, reducing treatment failures, and improving overall patient outcomes. However, current diagnostic tests present significant hurdles. Conventional phenotypic susceptibility testing relies largely on culture-based methods that are intrinsically slow and often delay reporting. Conversely, targeted molecular genotypic assays, often used in multiplex PCR panels, offer rapid turn-around times but remain limited to only a few resistance markers. To bridge this gap, whole-genome data derived from metagenomic sequencing has emerged as a powerful paradigm for comprehensive AMR phenotype prediction directly from clinical samples. Metagenomics enables the simultaneous detection of known resistance genes, point mutations, structural variations, and polygenic resistance markers without the delay of culture or the restrictions of targeted panels. This talk will provide an overview of the current landscape in genomic and metagenomic AMR prediction and outline key technological challenges. Furthermore, I will share our experience in this area.

X-linked genetic diseases affect sexes differently due to X-chromosome inactivation-driven differential methylation in females. Targeted long-read sequencing (LRS) via adaptive sampling (AS) characterizes genomic composition and methylation, but performs poorly with non-high-molecular-weight (non-HMW) DNA. We optimized an AS-based LRS protocol for non-HMW DNA and validated its diagnostic utility in 20 families with known X-linked defects previously identified by routine methods. Incorporating optimized DNA fragmentation, size selection, and library construction, our protocol achieved a >15-fold target enrichment compared to 8.1-fold using standard methods. The optimized LRS successfully detected all previously reported variants and uncovered novel genomic findings in 7/20 families (35%). By resolving structural variant breakpoints and determining gene methylation status, this protocol led to variant reclassification in 3/20 cases (15%). Ultimately, this optimized AS-based approach enables comprehensive genomic and methylation profiling of non-HMW DNA, offering a robust tool for evaluating X-linked defects with potential application to other disease states.

In this presentation, the translation of research-developed assays into laboratory-developed tests for pharmacogenomics (PGx) and CNS tumor classification through the clinical implementation of Oxford Nanopore sequencing in Thailand will be presented. The presentation will highlight how different implementation strategies are required for public and private hospitals, reflecting differences in clinical priorities, reimbursement models, laboratory infrastructure, and patient access. It will also discuss the multidisciplinary collaborations needed to bridge research, pathology, medical technologist, bioinformatics, and clinical care. Through real-world case studies, the audience will gain practical insights into assay development, validation, quality management, workflow integration, and service scale-up. This journey demonstrates how long-read sequencing can move beyond proof-of-concept to become a sustainable clinical service, delivering precision medicine that is both clinically impactful and accessible across diverse healthcare settings.

In clinical practice, we often encounter challenging cases in which contemporary techniques cannot always provide a complete diagnosis. This challenge is especially relevant to inherited disorders, for which the exact genetic aetiology may remain unknown even after a lengthy diagnostic odyssey. Recent advances in sequencing technologies have created new approaches to investigating genomic variation.

Driven by unmet diagnostic needs, we began developing and evaluating nanopore long-read sequencing capability at Princess Margaret Hospital. This session focuses on our experience with Flongle, Oxford Nanopore Technologies' compact, lower-cost flow-cell format, suited to focused, small-batch applications. Our long-read workflows characterised complex structural variants and evaluated repeat sizing and sequence interruptions in a proof-of-concept study of trinucleotide-repeat disorders. Ongoing analytical work extends to near-whole mitochondrial DNA sequencing for heteroplasmy measurement and large-deletion feasibility, and consensus-based ITS sequencing of survey and reference mushrooms for potential toxicology support.

Flongle's small-scale format provides a practical entry point for developing focused long-read applications. These examples demonstrate how we developed laboratory methods, bioinformatics workflows and interpretive experience with long-read sequencing. As experience, evidence and demand grow, the capability built through these projects may support broader applications at greater scale.

Alport syndrome (AS) is a prevalent inherited kidney disorder mainly caused by mutations in COL4A3, COL4A4, and COL4A5 genes. To elucidate the genetic variants of AS, we implemented a sequential sequencing strategy within a Chinese cohort of 555 patients, comprising whole-exome sequencing (WES) for all participants, followed by whole-genome sequencing (WGS), RNA sequencing (RNA-seq), and nanopore long-read sequencing (NLR-seq) for selected individuals. We identify 431 distinct variants in 509 (91.7%) patients, with 42.2% being novel. Beyond WES, additional sequencing approaches resolve 23 patients with noncoding, copy number, or structural variants. Notably, noncoding variants account for 16.2% of detected variants and exhibit ethnic-specific mutagenesis patterns. More interestingly, NLR-seq uncovers two novel types of structural variants, namely large insertions in intronic regions and complex duplication-inversion variants. This study provides deeper insights into the genetic architecture of AS and proposes a research paradigm for improving the genetic diagnosis of inherited diseases.

This study presents a selective sequencing platform capable of enriching target DNA regions, including CpG islands and flanking regions, achieving average enrichment of up to 1.7x, 40x in specific target region. By selectively excluding unwanted DNA during sequencing, the platform enhances efficiency and specificity. Leveraging deep learning models for signal classification, the platform demonstrates robust performance in remote processing on servers or in the cloud, reducing upfront costs and offering deployment flexibility. The results highlight the identification of significant differential methylated region associated with key transcripts such as dnm1p47, galnt9, mnx1-as1, and mnx1-as2, shedding light on potential regulatory mechanisms in cancer.

Nanopore sequencing continues to advance in accuracy, speed and breadth of analysis, bringing comprehensive genomic insights closer to routine clinical use. This presentation will highlight key technology updates and demonstrate how Oxford Nanopore sequencing of native DNA and RNA can resolve diverse genomic variation. Through emerging clinical applications – from rare disease and cancer to infectious disease and rapid, time-critical sequencing – we will explore how nanopore technology is turning complex genomic data into actionable answers.