WiPP

WiPP improves peak detection in large-scale gas chromatography-mass spectrometry (GC-MS) metabolomics datasets by integrating multiple peak picking algorithms and applying machine learning-based peak quality assessment.

Key Features:

• Multi-Algorithm Integration: Combines results from multiple commonly used peak detection algorithms to address variability in algorithmic approaches and parameter settings for GC-MS data.
• Machine Learning-Based Quality Assessment: Employs a machine learning classifier to evaluate detected peaks across seven distinct peak classes and assign quality scores.
• Automated Parameter Optimization: Automates optimization of algorithm parameters to improve accuracy and consistency of peak detection.
• Comprehensive Peak Set Creation: Merges quality information from individual peaks and results from multiple algorithms to produce a final high-quality peak set while retaining medium- and low-quality peaks for inspection.
• Impartial Performance Comparison: Standardizes evaluation to enable unbiased comparison of peak picking algorithm performance across diverse datasets.
• Implementation: Implemented in Python 3.

Scientific Applications:

• GC-MS metabolomics preprocessing: Produces optimized peak sets for downstream metabolomics analyses of GC-MS datasets.
• Algorithm benchmarking: Provides a standardized framework for comparing the performance of different peak picking algorithms.
• Analysis of complex samples and standards: Facilitates more robust and reproducible peak detection in complex biological samples and standard compound mixes.

Methodology:

Integrates multiple peak picking algorithms, evaluates detected peaks with a machine learning classifier based on seven predefined peak classes, automates parameter optimization, and combines these evaluations to generate an optimized comprehensive peak set while retaining medium- and low-quality peaks for further inspection.