metabCombiner

metabCombiner aligns and concatenates features from untargeted liquid chromatography–high resolution mass spectrometry (LC-HR-MS) datasets acquired under nonidentical conditions to enable cross-experimental metabolite matching and compound identification.

Key Features:

• Feature Matching Across Datasets: Aligns known and unknown metabolomic features between two untargeted LC-MS datasets to assemble a unified table of intersecting feature measurements.
• Handling Nonidentical Acquisition Conditions: Accounts for differing gradient elution methods and significant chromatographic retention time alterations between datasets.
• Grouping by Mass-to-Charge Ratio (m/z): Groups features by m/z values to create a search space for potential feature pair alignments.
• Retention Time Alignment (Spline Fitting): Fits a spline through selected retention time ordered pairs to map retention times across datasets.
• Ranking Alignments: Ranks candidate alignments using m/z similarity, mapped retention time, and relative abundance.
• Performance Evaluation: Demonstrated a mean absolute retention time prediction error of ~0.06 minutes and a weighted per-compound matching accuracy of ~90% on plasma metabolomics datasets.
• Versatility Across Biological Samples: Applied to plasma, urine, and muscle metabolomics datasets acquired from different laboratories.
• Support for Meta-Analyses and Collaborative Identification: Enables integration of disparate datasets to support collaborative compound identification and meta-analysis.

Scientific Applications:

• Cross-Study Compound Identification: Facilitates matching compound features across studies to improve annotation and identification confidence.
• Meta-Analysis of Metabolomics Data: Enables concatenation of measurements from heterogeneous LC-HR-MS experiments to support meta-analyses.
• Comparative Analysis Across Sample Types: Supports mapping and comparison of metabolite features in plasma, urine, and muscle datasets from different experimental conditions.

Methodology:

Features are grouped by m/z; a spline is fitted through selected retention time ordered pairs to align retention times across datasets; candidate alignments are ranked by m/z similarity, mapped retention time, and relative abundance.