Workflow for improved Peak Picking (WiPP) introduces a novel approach to automatic peak picking in GC-MS data in order to optimise the accuracy and quality of the process by combining the strengths of multiple existing or new peak picking algorithms.
We apply a visualisation strategy combined with a support vector machine learning approach to automatically assess peak quality. The generated model is used for two or more peak picking algorithms whose results can then themselves be scored for quality. The results of this scoring allows the integration of peak detection algorithm outputs in a final high quality dataset which maximises peak number while minimising false peak discovery.
WiPP offers three key improvements to existing GC-MS peak detection tools:
- Optimal use of individual peak picking algorithms by optimizing their parameters.
- Larger peak coverage by combining peaks detected by different algorithms.
- Reduction of reported false positive peaks using automated machine learning classification.
WiPP significantly reduces the “hands on” time spent on the analysis of large scale GC-MS metabolomics data. WiPP is an extendable modular pipeline and will be made publicly available as an open source software on GitHub.
Last modified: Jan 1, 0001