Prediction of pyrazinamide resistance in Mycobacterium tuberculosis using structure-based machine learning approaches

Background: Pyrazinamide is one of four first-line antibiotics used to treat tuberculosis, however antibiotic susceptibility testing for pyrazinamide is challenging. Resistance to pyrazinamide is primarily driven by genetic variation in pncA, an enzyme that converts pyrazinamide into its active form...

Full description

Saved in:
Bibliographic Details
Published in:bioRxiv 2023-11
Main Authors: Carter, Joshua, Walker, Timothy M, Walker, A Sarah, Whitfield, Michael, Morlock, Glenn P, Lynch, Charlotte I, Adlard, Dylan, Peto, Timothy Ea, Posey, Jamie E, Crook, Derrick W, Fowler, Philip W
Format: Article
Language:English
Subjects:
Antibiotics
Artificial intelligence
Clinical isolates
Clinical microbiology
Genetic diversity
Learning algorithms
Machine learning
Missense mutation
Mutation
Phenotypes
Pyrazinamide
Tuberculosis
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Background: Pyrazinamide is one of four first-line antibiotics used to treat tuberculosis, however antibiotic susceptibility testing for pyrazinamide is challenging. Resistance to pyrazinamide is primarily driven by genetic variation in pncA, an enzyme that converts pyrazinamide into its active form. Methods: We curated a dataset of 664 non-redundant, missense amino acid mutations in pncA with associated high-confidence phenotypes from published studies and then trained three different machine learning models to predict pyrazinamide resistance. All models had access to a range of protein structural-, chemical- and sequence-based features. Results: The best model, a gradient-boosted decision tree, achieved a sensitivity of 80.2% and a specificity of 76.9% on the hold-out Test dataset. The clinical performance of the models was then estimated by predicting the binary pyrazinamide resistance phenotype of 4,027 samples harboring 367 unique missense mutations in pncA derived from 24,231 clinical isolates. Conclusions: This work demonstrates how machine learning can enhance the sensitivity/specificity of pyrazinamide resistance prediction in genetics-based clinical microbiology workflows, highlights novel mutations for future biochemical investigation, and is a proof of concept for using this approach in other drugs.Competing Interest StatementThe authors have declared no competing interest.Footnotes* The last version was preprinted in 2019; we picked up the manuscript again after the pandemic and have included more samples for training. Another important change is that all the code (from original datasets, data cleaning, test/train split, training ML models, analysis etc) is now publicly available on GitHub. Hence this manuscript is significantly improved although it reaches a similar conclusion.* https://github.com/fowler-lab/predict-pyrazinamide-resistance
DOI:10.1101/518142