IPO

IPO optimizes parameter settings for untargeted LC-HRMS metabolomics data processing to improve peak detection, retention time alignment, and peak grouping accuracy.

Key Features:

• Label-free optimization: Uses natural stable 13C isotopic peaks rather than experimental labeling to derive scoring for parameter selection.
• Applicability across LC-HRMS platforms: Applicable to data from liquid chromatography–high-resolution mass spectrometry (LC-HRMS) across different sample types and chromatographic techniques.
• XCMS parameter optimization: Optimizes XCMS peak-picking, retention time correction, and grouping parameters.
• Peak picking score calculation: Calculates a peak picking score based on natural stable 13C isotopic peaks.
• Retention time correction optimization: Minimizes relative retention time differences within peak groups to improve temporal alignment.
• Grouping parameter optimization: Maximizes the number of peak groups that contain one peak from each injection of a pooled sample.
• Design of Experiments (DoE): Employs DoE to systematically explore combinations of parameter settings.
• Response surface models: Uses response surface models to evaluate scores from different parameter configurations and identify optima.
• Implementation: Implemented in R for computational execution of the optimization procedures.

Scientific Applications:

• Improved grouping reliability: Increased the number of reliable groups by 146% to 361% in reported evaluations.
• Reduced non-reliable groups: Decreased occurrences of non-reliable groups by 3% to 8%.
• Improved retention time alignment: Reduced retention time deviation to approximately one-third of prior values.
• Validation on multiple datasets: Demonstrated performance across three distinct datasets each comprising a training set and a test set.

Methodology:

Calculates a peak-picking score using natural stable 13C isotopic peaks to optimize XCMS parameters, minimizes relative retention time differences within peak groups, maximizes peak groups with one peak per pooled-sample injection, employs Design of Experiments to explore parameter settings, and evaluates parameter sets with response surface models.