Antibiotic resistance (ABR) is a major threat to human health worldwide, with increasing instances of multi-drug resistant pathogens emerging to diminish the effectiveness of antibiotics(1). Whole-genome sequencing (WGS) is rapidly changing the clinical microbiology landscape, with exciting potential for rapidly and accurately detecting ABR in diagnostic laboratories, a crucial factor in infection control and treatment. Most work to date has focussed on the development of software capable of detecting the presence of mobile genetic elements conferring ABR from WGS data(2). However, less consideration has been given to the identification of chromosomally-encoded ABR mechanisms, such as single-nucleotide polymorphisms (SNPs), insertion-deletions (indels), copy number variants (CNV), and functional gene loss.
We present an improved software for Antibiotic Resistance Detection and Prediction (ARDaP) from WGS data. ARDaP was designed with two main aims: 1) to accurately identify all characterised ABR genetic mechanisms and present the predicted ABR profile in an easy-to-interpret report; and 2) to predict enigmatic ABR mechanisms based on i) novel mutants in known ABR-conferring genes, or ii) a microbial genome-wide association approach that correlates ABR phenotypes with genetic variants to identify putative causative mutant/s.
We demonstrate the applicability of ARDaP using the Tier 1 select agent and melioidosis pathogen, Burkholderia pseudomallei, as a model organism due to its exclusively chromosomally-encoded ABR mechanisms and high mortality rate (3). Using an extensive, well-characterised collection of 991 B. pseudomallei clinical strains, we demonstrate that ARDaP can accurately detect all known ABR mechanisms in B. pseudomallei (>40 mutations) with high rates of precision and recall. Furthermore, ARDaP predicted four novel loss-of-function mutations that decreased meropenem susceptibility in B. pseudomallei; this phenotype is associated with increased treatment failure and fatality rates(3). ARDaP is a comprehensive and accurate tool for identifying and predicting ABR mechanisms from WGS data. Its clinician-friendly report(4), which summarises a given strain’s AbR profile, holds great promise for informing personalised treatment regimens and treatment shifts in response to the detection of precursor or ABR-conferring mutations.