Journal Article

MALDI-TOF mass spectrometry on intact bacteria combined with a refined analysis framework allows accurate classification of MSSA and MRSA

Larrouy-Maumus, Gerald, Ranganathan, Nisha, Tang, Wenhao, Shahrezaei, Vahid

2019

Overview

Fast and reliable detection coupled with accurate data-processing and analysis of antibiotic-resistant bacteria is essential in clinical settings. In this study, we use MALDI-TOF on intact cells combined with a refined analysis framework to demonstrate discrimination between methicillin-susceptible (MSSA) and methicillin-resistant (MRSA) Staphylococcus aureus. By combining supervised and unsupervised machine learning methods, we firstly show that the mass spectroscopy data contains strong signal for the clustering of MSSA and MRSA. Then we concentrate on applying supervised learning to extract and verify the important features. A new workflow is proposed that allows for extracting a fixed set of reference peaks so that any new data can be aligned to it and hence consistent feature matrices can be obtained. Also note that by doing so we are able to examine the robustness of the important features that have been found. We also show that appropriate size of the benchmark data, appropriate alignment of the testing data and use of an optimal set of features via feature selection results in prediction accuracy over 90%. In summary, as proof-of-principle, our integrated experimental and bioinformatics study suggests a novel intact cell MALDI-TOF to be of great promise for fast and reliable detection of MRSA strains.

• Publisher: Public Library of Science,Public Library of Science (PLoS)
• Subject: Amino Acid Sequence / Anti-Bacterial Agents - pharmacology / Antibiotic resistance / Antibiotics / Artificial intelligence / Bacteria / Bacterial Proteins - analysis / Bacterial Proteins - chemistry / Bacteriology / Benchmarking / Bioinformatics / Biology and life sciences / Biomarkers / Cancer / Classification / Clustering / Computational Biology / Computer and Information Sciences / Data processing / Datasets / Discriminant analysis / Dosage and administration / Drug resistance / Engineering and Technology / Feature extraction / Humans / Identification / Ions / Laboratories / Lasers / Learning algorithms / Machine learning / Mass spectrometry / Mass spectroscopy / Medical research / Medicine and health sciences / Methicillin / Methicillin - pharmacology / Methicillin-Resistant Staphylococcus aureus - chemistry / Methicillin-Resistant Staphylococcus aureus - classification / Methicillin-Resistant Staphylococcus aureus - drug effects / Methods / Microbial drug resistance / Microbiology / Novels / Pathogens / Peptide Mapping - methods / Peptide Mapping - statistics & numerical data / Peptides / Physical Sciences / Proteins / Proteomics / Research and Analysis Methods / Scientific imaging / Spectrometry, Mass, Matrix-Assisted Laser Desorption-Ionization - methods / Spectroscopy / Staphylococcus aureus / Staphylococcus aureus - chemistry / Staphylococcus aureus - classification / Staphylococcus aureus - drug effects / Staphylococcus aureus infections / Staphylococcus infections / Supervised Machine Learning / Support Vector Machine / Values / Workflow / Workflow software