PeakDecoder enables machine learning-based metabolite annotation and accurate profiling in multidimensional mass...

Area of Science:

• Biochemistry
• Analytical Chemistry
• Computational Biology

Background:

• Untargeted metabolomics offers insights into biological and environmental processes.
• Challenges exist in analyzing complex data from multidimensional measurements due to a lack of rapid methods and robust algorithms.

Purpose of the Study:

• To develop and evaluate a sensitive, high-throughput analytical and computational workflow for accurate metabolite profiling.
• To address limitations in analyzing heterogeneous data from advanced separation and mass spectrometry techniques.

Main Methods:

• Combined liquid chromatography, ion mobility spectrometry, and data-independent acquisition mass spectrometry.
• Utilized PeakDecoder, a machine learning algorithm, to differentiate true co-elution/co-mobility and calculate metabolite identification error rates.
• Applied the workflow to engineered strains of Aspergillus pseudoterreus, Aspergillus niger, Pseudomonas putida, and Rhodosporidium toruloides.

Main Results:

• Successfully applied the PeakDecoder algorithm for metabolite profiling across 116 microbial sample runs.
• Confidently annotated and quantified 2683 features using a library of 64 standards.
• Validated results against selected reaction monitoring and gas-chromatography platforms.

Conclusions:

• The developed workflow enables sensitive and high-throughput metabolite profiling.
• PeakDecoder significantly improves the accuracy and efficiency of analyzing complex metabolomics data.
• This approach facilitates a deeper understanding of biochemical processes in various microbial systems.