Shape-based feature matching improves protein identification via LC-MS and tandem MS.

Area of Science:

• Proteomics and Mass Spectrometry
• Bioinformatics and Computational Biology

Background:

• Liquid chromatography-mass spectrometry (LC-MS) is crucial for protein characterization but faces challenges with dynamic range and complexity.
• Inconsistent analyte travel times in LC columns cause nonlinear retention time (RT) shifts, complicating peptide feature matching across samples.
• Existing RT alignment methods are computationally expensive or struggle with ambiguities in feature matching.

Purpose of the Study:

• To develop a novel feature matching algorithm incorporating wavelet-based shape information for improved LC-MS data alignment.
• To create a nonlinear feature-based alignment framework for LC-MS experiments by combining the new algorithm with robust nonparametric kernel-type regression.
• To evaluate the algorithm's performance in both LC-MS and MALDI MS/MS applications for enhanced peptide identification and data reproducibility.

Main Methods:

• A novel feature matching algorithm utilizing wavelet-based shape information was developed.
• The algorithm was integrated with nonparametric kernel-type regression to create a nonlinear alignment framework for LC-MS data.
• The framework was validated on complex LC-MS samples with spiked-in proteins and on MALDI MS/MS data.

Main Results:

• The developed alignment framework demonstrated higher reproducibility and probability scores for identifying spiked-in proteins compared to SuperHirn software.
• The framework increased the number of matched features and improved correlation between replicates in LC-MS experiments.
• Application to MALDI MS/MS data showed improved peptide identification by utilizing matched features across tandem mass spectra replicates.

Conclusions:

• The proposed feature matching algorithm and alignment framework effectively address nonlinear RT shifts in LC-MS.
• The method enhances the accuracy, reproducibility, and scope of peptide identification in complex proteomic datasets.
• This approach offers a significant improvement over existing methods for both LC-MS and MALDI MS/MS data analysis.