Generalized precursor prediction boosts identification rates and accuracy in mass spectrometry based proteomics

Area of Science:

• Proteomics
• Biomarker Discovery
• Mass Spectrometry

Background:

• Data independent acquisition mass spectrometry (DIA-MS) is valuable for blood-based biomarker identification.
• Large search spaces in plasma proteome analysis can lead to false positives and compromise false discovery rate (FDR) accuracy.

Purpose of the Study:

• To develop a novel method for controlling FDR and increasing protein identification in DIA-MS.
• To improve the accuracy and generalizability of DIA-MS for biomarker discovery.

Main Methods:

• Developed a generalized precursor scoring (GPS) method trained on 2.75 million precursors.
• Applied GPS to DIA-MS data for plasma proteome analysis.
• Validated GPS performance on new datasets and for biomarker identification.

Main Results:

• GPS confidently controls FDR while increasing protein identification rates in DIA-MS.
• GPS demonstrates generalizability to new datasets and improves quantitative accuracy.
• A panel of blood-based protein biomarkers for septic acute kidney injury subphenotypes was identified using GPS.

Conclusions:

• The generalized precursor scoring (GPS) method enhances DIA-MS performance for biomarker discovery.
• GPS offers a robust approach to control FDR and increase protein identification in complex proteomic samples.
• GPS facilitates accurate identification of blood-based biomarkers, as demonstrated in septic acute kidney injury subphenotype discrimination.