Fast Pathogen Identification Using Single-Cell Matrix-Assisted Laser Desorption/Ionization-Aerosol Time-of-Flight...

Area of Science:

• Microbiology
• Analytical Chemistry
• Bioinformatics

Background:

• Matrix-assisted laser desorption/ionization-time-of-flight mass spectrometry (MALDI-TOF MS) is crucial for identifying pathogenic microorganisms in infectious disease diagnostics.
• Current MALDI-TOF MS protocols require significant biomass, necessitating time-consuming bacterial culturing and amplification steps.
• This pre-analytical step limits the speed of microbial identification.

Purpose of the Study:

• To investigate the feasibility of direct MALDI-TOF MS analysis on individual bacterial cells, eliminating the need for culturing.
• To develop and evaluate a deep learning architecture for analyzing single-cell MALDI-TOF MS data.
• To compare the performance of deep learning with traditional machine learning algorithms for rapid species identification.

Main Methods:

• Direct MALDI-TOF MS analysis was performed on individual bacterial cells, bypassing the traditional culturing step.
• A novel deep learning architecture was proposed for the analysis of the generated spectral data.
• The deep learning model's performance was benchmarked against established supervised machine learning algorithms.
• The workflow was validated on a comprehensive dataset of bacterial species commonly associated with urinary tract infections.

Main Results:

• Direct MALDI-TOF MS analysis of individual bacterial cells enables faster species identification compared to culture-based methods.
• The proposed deep learning architecture achieved high accuracy in discriminating between different bacterial species.
• Accuracies of up to 85% were obtained for discriminating five distinct bacterial species using the deep learning approach.
• The study successfully demonstrated the potential of single-cell MALDI-TOF MS coupled with deep learning for rapid diagnostics.

Conclusions:

• Culturing can be omitted for MALDI-TOF MS-based microbial identification, significantly accelerating the diagnostic process.
• Deep learning models offer a powerful and accurate method for analyzing single-cell MALDI-TOF MS data.
• This approach holds promise for rapid and reliable diagnostics of infectious diseases, particularly in resource-limited settings or time-critical situations.