Identification of Mycobacterium abscessus Subspecies by MALDI-TOF Mass Spectrometry and Machine Learning

Area of Science:

• Microbiology
• Computational Biology
• Spectrometry

Background:

• Mycobacterium abscessus is a common and pathogenic nontuberculous mycobacteria (NTM).
• It comprises three subspecies: M. abscessus subsp. abscessus, M. abscessus subsp. bolletii, and M. abscessus subsp. massiliense.
• Accurate differentiation is crucial due to varying antibiotic susceptibility.

Purpose of the Study:

• To develop a rapid and accurate method for differentiating M. abscessus subspecies.
• To leverage MALDI-TOF MS protein spectra with machine learning for subspecies identification.

Main Methods:

• Analysis of 325 clinical M. abscessus isolates using MALDI-TOF MS.
• Development of machine learning predictive models (random forest) based on protein spectra.
• Evaluation of factors like culture media, colony morphology, and geographic origin.

Main Results:

• A random forest model achieved 96.5% accuracy in subspecies identification with confidence criteria.
• An improved model identified 88.9% of international strains.
• Geographic origin significantly impacted protein spectra; two novel differentiating peaks were identified.

Conclusions:

• Machine learning applied to MALDI-TOF MS data is a promising tool for rapid and accurate M. abscessus subspecies identification.
• This method can improve clinical diagnosis and treatment strategies for NTM infections.
• Further research into novel protein peaks may enhance differentiation capabilities.