TIMS²Rescore: A Data Dependent Acquisition-Parallel Accumulation and Serial Fragmentation-Optimized Data-Driven...

Area of Science:

• Proteomics and Mass Spectrometry
• Bioinformatics and Computational Biology
• Biotechnology and Instrumentation

Background:

• High-throughput mass spectrometry (MS) is crucial for biological and disease research.
• Specialized proteomics applications face challenges like increased identification ambiguity.
• Advancements in MS instrumentation and AI are needed to improve data analysis.

Purpose of the Study:

• To introduce TIMS²Rescore, a novel data-driven rescoring workflow for timsTOF MS data.
• To address identification ambiguity in complex proteomics datasets.
• To streamline data analysis for plasma proteomics, immunopeptidomics, and metaproteomics.

Main Methods:

• Development of TIMS²Rescore, a workflow optimized for DDA-PASEF data.
• Incorporation of new timsTOF MS²PIP spectrum prediction models.
• Integration of IM2Deep, a deep learning-based peptide ion mobility predictor.
• Direct acceptance of raw mass spectrometry data and search results from various engines.

Main Results:

• TIMS²Rescore effectively handles DDA-PASEF data from timsTOF instruments.
• The workflow demonstrates robust performance on diverse datasets including plasma, immunopeptidomics, and metaproteomics.
• AI-driven predictions and ion mobility data integration enhance identification accuracy.

Conclusions:

• TIMS²Rescore offers a powerful solution for improving protein identification in complex MS data.
• The open-source platform facilitates broader adoption and advancement in proteomics research.
• This workflow significantly contributes to overcoming analytical challenges in specialized proteomics fields.