Normalization of peak intensities in bottom-up MS-based proteomics using singular value decomposition.

Area of Science:

• Proteomics
• Biotechnology
• Computational Biology

Background:

• Liquid chromatography-mass spectrometry (LC-MS) is crucial for protein identification and quantification.
• High-throughput LC-MS data often contain systematic biases, necessitating robust normalization methods.
• Effective normalization enhances accuracy and precision in comparative proteomics.

Purpose of the Study:

• To develop a flexible and accurate normalization algorithm for LC-MS data.
• To address challenges of widespread missing measurements and prevent overfitting in bias correction.
• To integrate a novel normalization approach into existing proteomics analysis pipelines.

Main Methods:

• Proposed EigenMS algorithm, an adaptation of surrogate variable analysis (SVA).
• Utilizes singular value decomposition (SVD) to capture and remove biases.
• Incorporates specific adaptations for handling missing data and preventing overfitting.

Main Results:

• EigenMS effectively captures and removes systematic biases from LC-MS peak intensity measurements.
• Demonstrated performance using large-scale calibration data and simulations.
• Outperforms existing normalization alternatives in accuracy and precision.

Conclusions:

• EigenMS provides a robust solution for normalizing complex LC-MS data.
• The algorithm improves the reliability of quantitative proteomics.
• Software is available in the DAnTE platform and as standalone applications.