Biology subjects

Vandelannoote, K.

Publications and source records attributed to Vandelannoote, K..

3 recordsLinked to original sources

bioRxiv2024

hAMRonization: Enhancing antimicrobial resistance prediction using the PHA4GE AMR detection specification and tooling

The detection of antimicrobial resistance (AMR) markers directly from genomic or metagenomic data is becoming a standard clinical and public health procedure. This has resulted in the development of a number of different bioinformatic AMR prediction tools. Although many may implement similar principles, these tools differ significantly in their supported inputs, search algorithms, parameterisation, and underlying reference databases. Each of these tools generates a report of detected AMR genes or variants in a distinct, non-standard, format. This presents a huge barrier to the comparison of results and to the modularity of tools for AMR gene prediction within bioinformatic workflows. In collaboration with 17 public health laboratories across 10 countries, the Public Health Alliance for Genomic Epidemiology (PHA4GE) (https://pha4ge.org) data structures working group has developed and piloted a standardized output specification for the bioinformatic detection of AMR from microbial genomes. In this report, we discuss hAMRonization, a python package and command-line utility, which implements PHA4GEs AMR specification to combine the outputs of disparate antimicrobial resistance gene detection tools into a single unified format. hAMRonization can be easily extended and currently supports 18 different tools (both species-agnostic and species-specific) for the detection of genes and/or variants conferring AMR. The harmonized reports are available in tabular form, JSON format or through an interactive HTML file (e.g., https://maguire-lab.github.io/assets/interactive_report_demo.html) that can be opened within the browser for navigable data exploration. As of 2024-03-07 hAMRonization has been downloaded [~]12,500 times, incorporated into >9 public bioinformatic tools and workflows, and been internally adopted by several national and international public health groups. The hAMRonization tool and underlying specification are open-source and freely available through PyPI, conda and GitHub (https://github.com/pha4ge/hAMRonization).

bioinformatics↗

bioRxiv2023

High performance Legionella pneumophila source attribution using genomics-based machine learning classification

Fundamental to effective Legionnaires disease outbreak control is the ability to rapidly identify the environmental source(s) of the causative agent, Legionella pneumophila. Genomics has revolutionised pathogen surveillance but L. pneumophila has a complex ecology and population structure that can limit source inference based on standard core genome phylogenetics. Here we present a powerful machine learning approach that assigns the geographical source of Legionnaires disease outbreaks more accurately than current core genome comparisons. Models were developed upon 534 L. pneumophila genome sequences, including 149 genomes linked to 20 previously reported Legionnaires disease outbreaks through detailed case investigations. Our classification models were developed in a cross-validation framework using only environmental L. pneumophila genomes. Assignments of clinical isolate geographic origins demonstrated high predictive sensitivity and specificity of the models, with no false positives or false negatives for 13 out of 20 outbreak groups, despite the presence of within-outbreak polyclonal population structure. Analysis of the same 534-genome panel with a conventional phylogenomic tree and a core genome multi-locus sequence type allelic distance-based classification approach revealed that our machine learning method had the highest overall classification performance - agreement with epidemiological information. Our multivariate statistical learning approach maximises use of genomic variation data and is thus well-suited for supporting Legionnaires disease outbreak investigations.

bioinformatics↗

bioRxiv2022

Structured surveys of Australian native possum excreta predict Buruli ulcer occurrence in humans

Buruli ulcer (BU) is a neglected tropical disease caused by infection of subcutaneous tissue with Mycobacterium ulcerans. BU is commonly reported across rural regions of Central and West Africa but has been increasing dramatically in temperate southeast Australia around the major metropolitan city of Melbourne. Previous research has shown that Australian native possums are reservoirs of M. ulcerans and that they shed the bacteria in their fecal material (excreta). Field surveys show that locales where possums harbor M. ulcerans overlap with human cases of BU, raising the possibility of using possum excreta surveys to predict the risk of disease occurrence in humans. We thus established a highly structured 12-month possum excreta surveillance program across an area of 350 km2 in the Mornington Peninsula area 70 km south of Melbourne, Australia. The primary objective of our study was to assess if M. ulcerans surveillance of possum excreta provided useful information for predicting future human BU case locations. Over two sampling campaigns in summer and winter, we collected 2282 possum excreta specimens of which 11% were PCR positive for M. ulcerans-specific DNA. Using the spatial scanning statistical tool SatScan, we observed non-random, co-correlated clustering of both M. ulcerans positive possum excreta and human BU cases. We next trained a statistical model with the Mornington Peninsula excreta survey data to predict the future likelihood of human BU cases occurring in the region. By observing where human BU cases subsequently occurred, we show that the excreta model performance was superior to a null model trained using the previous years human BU case incidence data (AUC 0.66 vs 0.55). We then used data unseen by the excreta-informed model from a new survey of 661 possum excreta specimens in Geelong, a geographically separate BU endemic area to the southwest of Melbourne, to prospectively predict the location of human BU cases in that region. As for the Mornington Peninsula, the excreta-based BU prediction model outperformed the null model (AUC 0.75 vs 0.50) and pinpointed specific locations in Geelong where interventions could be deployed to interrupt disease spread. This study highlights the One Health nature of BU by confirming a quantitative relationship between possum excreta shedding of M. ulcerans and humans developing BU. The excreta survey-informed modeling we have described will be a powerful tool for efficient targeting of public health responses to stop BU.

microbiology↗