Peptidome-Driven Algorithms for Faster and More Precise Mycobacterial Detection - Abstract The global incidence of nontuberculous mycobacteria (NTM) infections is rising, posing significant public health challenges, particularly in developed countries. Rapid and accurate NTM species identification and drug susceptibility predictions are critical, as treatment varies by species, and delays or inaccuracies can lead to ineffective therapies, drug toxicity, and the emergence of drug-resistant (DR) NTM strains. Current diagnostic methods fail to address these challenges effectively. Proteins secreted by mycobacteria, although conserved, harbor peptide sequence variants that can uniquely identify closely related pathogens and serve as diagnostic biomarkers. Mass spectrometry (MS) with algorithm assistance enables the detection of these variants, offering a platform for precise species identification and improved drug susceptibility prediction. In our previous studies, we have systematically evaluated the power of MS for mycobacteria identification using peptides. We thus propose that a refined LC-MS/MS assay approach could provide comprehensive peptidomics information required to improve NTM species identification and drug susceptibility predictions. As a proof of concept, we have created a computational algorithm-based MS assay (MycoID) for NTM identification and have demonstrated that this approach can accurately identify NTM species and subspecies. This proposal will refine and evaluate the assay in identifying NTM species and subspecies, and predicting drug resistance, by analyzing peptide sequence variants and changes in peptide abundance, to provide information not available from genomics data. This project aims to refine and expand the MycoID platform to achieve three specific goals: 1. Enhancing NTM pipeline: Develop a custom MycoPep library to improve the specificity and accuracy of NTM species and subspecies identification, leveraging in-silico peptide libraries and optimizing the MycoID pipeline. 2. Assess the performance of MycoID to identify species and subspecies: Evaluate MycoID’s performance on diverse mass spectrometry instruments, including clinically affordable systems, and characterize co-infections and early detection potential in a large NIH NTM cohort. 3. Evaluate the capacity of the MycoID platform: Integrate a refined Machine Learning-based model into the MycoID pipeline to predict macrolide resistance in Mycobacterium abscessus isolates, utilizing a retrospective NIH cohort for validation and focusing on macrolide resistance as a critical clinical need. This integrative approach will transform the diagnostic landscape by reducing the time to species identification and drug susceptibility testing while enabling continuous platform refinements for clinical and research applications. If successful, the MycoID platform will significantly enhance the accuracy and timeliness of NTM diagnosis and treatment.
