Purdue University Graduate School
Browse

Leveraging of Machine Learning to Evaluate Genotypic-Phenotypic Concordance of Pasteurella Multocida Isolated from Bovine Respiratory Disease Cases

Download (1.12 MB)
thesis
posted on 2023-04-27, 20:56 authored by Tessa R SheetsTessa R Sheets

 Pasteurella multocida is a respiratory pathogen that is frequently isolated from cattle suffering  from bovine respiratory disease (BRD), the leading cause of mortality and morbidity on modern day cattle farms. Treatment involves the use of antimicrobials which have been shown to fail for  about 30% of BRD cases, leading to the suspicion that etiologic agents, such as P. multocida, may  be resistant. Phenotypic resistance can be confirmed via laboratory antibiotic susceptibility testing  (AST) but this requires several days to complete. Genotypic resistance could be quickly assessed  via nucleic acid assays based on the presence of known antibiotic resistance genes (ARGs). In  human medicine, resistant genes associated with common antibiotics (i.e., ampicillin and penicillin)  in common pathogens (i.e., Salmonella) are very accurate in predicting phenotypic resistance;  however, ARGs associated with antibiotics used to treat BRD, such as enrofloxacin and  tulathromycin, have shown low genotype-phenotype concordance. Hence, this study aims to  improve P. multocida genotype-phenotype concordance by applying a machine learning (ML)  algorithm to identify novel genomic sequences (biomarkers) that have greater accuracy in  predicting resistance to antibiotics commonly used to treat BRD compared to known ARGs.  Cultures of P. multocida were isolated from cattle with clinical signs of BRD. Antibiotic  susceptibility testing was performed and recorded for each isolate. Genomes were sequenced and  assembled, followed by annotating and identifying ARGs using the comprehensive antibiotic  resistance database (CARD). Assembled genomes were then split into 31-base long segments (31- mers), and these segments along with phenotypic antibiotic susceptibility were used as input data  for the ML algorithm. Important genomic biomarkers for four out of the six tested antibiotics were  found to have greater accuracy when predicting resistance phenotype compared to known ARGs.  The biomarker for enrofloxacin had the highest accuracy of 100% whereas the accuracy for the  12 tulathromycin biomarker was 81% but was still greater than the accuracy given by ARGs of 63%.  On the other hand, resistance genes for florfenicol and tetracycline showed greater genotype?phenotype concordance, with accuracies of 95% and 91%, respectively. Annotations to important  rulesets determined by ML were associated with clustered regularly interspaced short palindromic  repeats (CRISPR) sequences, ligases that function to recycle murein into the peptidoglycan (PDG) layer, and transferases that control the synthesis and modulation of the lipopolysaccharide (LPS).  External validation revealed that phenotypic resistance could be accurately predicted for  danofloxacin and enrofloxacin using genomic biomarkers determined by ML, and for florfenicol  using the floR gene. This study demonstrated that genomic biomarkers determined by ML can provide an accurate prediction of antibiotic resistance within Pasteurella multocida isolates.  Assays could be developed to target ML-generated biomarkers and known ARGs to predict resistance in sick animals and to limit treatment failures associated with antibiotic resistance in  cattle suffering from BRD. 

Funding

Foundation for Food and Agriculture Research – Grant ID: ICASATWG-0000000022

History

Degree Type

  • Master of Science

Department

  • Animal Sciences

Campus location

  • West Lafayette

Advisor/Supervisor/Committee Chair

Dr. Timothy Johnson

Additional Committee Member 2

Dr. Michael Gribskov

Additional Committee Member 3

Dr. Deepti Pillai

Usage metrics

    Licence

    Exports

    RefWorks
    BibTeX
    Ref. manager
    Endnote
    DataCite
    NLM
    DC