Multi-Reference Spectral Library Yields Almost Complete Coverage of Heterogeneous LC-MS/MS Data Sets

Area of Science:

• Proteomics
• Analytical Chemistry
• Bioinformatics

Background:

• Spectral libraries are crucial for data-independent-acquisition (DIA) proteomics.
• Current methods assume a single characteristic intensity pattern (CIP) per peptide charge pair, which is often insufficient.
• Significant spectral variability exists even under consistent experimental conditions.

Purpose of the Study:

• To systematically evaluate spectral variability in DIA proteomics.
• To develop an improved spectral library approach addressing spectral variability.
• To enhance peptide identification rates in DIA experiments.

Main Methods:

• Systematic evaluation of spectral variability across public and in-house datasets.
• Clustering of preprocessed spectra to derive multiple characteristic intensity patterns (MCIPs) for each peptide charge pair.
• Comparison of MCIP libraries derived from public repositories versus custom-made libraries.

Main Results:

• Widespread spectral variability was confirmed, occurring even under fixed experimental conditions.
• MCIPs provide near-complete coverage of heterogeneous spectral data without increasing false discovery rates.
• MCIP libraries from public repositories perform comparably to custom-made libraries.
• Application of the MCIP approach significantly increased peptide recognition in a DIA dataset.

Conclusions:

• The assumption of a single CIP is a limitation in current DIA proteomics spectral library tools.
• The MCIP approach effectively captures spectral variability and improves peptide identification.
• MCIPs offer an easily implementable enhancement for spectral library search engines and better utilization of spectral repositories.