Development of MALDI-TOF MS based discrimination of bacterial strains within species for fast prediction of antibiotic resistance and for efficient prevention of nosocomial transmission.

01 November 2020 → Ongoing
Research Foundation - Flanders (FWO)
Research disciplines
  • Medical and health sciences
    • Bioinformatics of disease
    • Computational biomodelling and machine learning
    • Development of bioinformatics software, tools and databases
    • Bacteriology
    • Infectious diseases
Matrix-assisted laser-desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS or MT) typing antibiotic resistance machine learning whole-genome sequencing hospital hygiene management Pseudomonas aeruginosa
Project description

Application of MALDI-TOF (MT) technology revolutionized the field of diagnostic microbiology for the identification of bacterial isolates cultured from clinical samples. In a few seconds, MT generates complex peptide spectra from isolates and compares a simplified spectrum to a database of spectra from known species to obtain an identification. However, the most important information for the clinician, i.e. the susceptibility of the isolate to different antibiotics, still depends on the antibiogram and takes 1 to 3 days, to be obtained. Meanwhile, patients are treated empirically and prescribed antibiotics that are too often inappropriate. So, in case clinical laboratories could rely on rapid, reliable and cost-effective methods not only for the identification of bacteria from clinical samples, but also to predict more quickly antibiotic susceptibility, appropriate antibiotic therapy might be initiated early. This proposal aims at exploiting all data generated by MS to not only identify isolates to the species level, but to immediately recognize individual bacterial strains and as such evaluate to what extent strain identification enables accurate antibiotic susceptibility prediction. Protein extractions will be obtained by ethanol/formic acid extraction. The MT-acquired peptide spectra will be analyzed by the implementation of machine learning algorithms for the identification of the isolate and to construct predictive models for the antibiotic susceptibility of the strain.