An m/z- and Intensity-Based HRMS Clustering Algorithm Targeted toward Single-Cell Metabolomics.

Area of Science:

• Analytical Chemistry
• Biochemistry
• Cell Biology

Background:

• Single-cell (SC) metabolomics uses high-resolution mass spectrometry (HRMS) to analyze cellular components, offering potential for diagnostics and mechanistic insights.
• Existing HRMS data processing software is not optimized for SC HRMS data, lacking necessary features for across-scan clustering and across-sample alignment of m/z values.

Purpose of the Study:

• To present a novel, robust algorithm for SC HRMS m/z clustering and alignment.
• To compare the performance of the new algorithm against existing commercial and open-source software.

Main Methods:

• Development of the Global Clustering unTargeted Analysis (GCTA) algorithm, featuring strict cluster size limits and noise filtering.
• Comparison of GCTA with Sciex MarkerView, Thermo FreeStyle, DBSCAN, and MaldiQuant binning using standard mixes and SC HRMS data from two cell lines.
• Evaluation based on accuracy in identifying known m/z values from databases (HMDB, LipidMaps) and standard compounds.

Main Results:

• GCTA demonstrates accuracy comparable to commercial software (MarkerView, FreeStyle) for m/z clustering and alignment.
• The algorithm effectively filters noise and is less prone to peak splitting than MaldiQuant binning.
• GCTA shows competitive performance against DBSCAN, MaldiQuant binning, and MarkerView when compared to theoretical m/z values.

Conclusions:

• The developed GCTA algorithm is a pivotal tool for untargeted SC HRMS metabolomics, providing reliable compound identification and data processing.
• GCTA's ability to perform both m/z clustering and reference-free alignment enhances its utility for advancing SC metabolomics research.