Data dependent-independent acquisition proteomics (DDIA)

Data dependent-independent acquisition proteomics (DDIA) integrates data dependent acquisition (DDA) and data independent acquisition (DIA) within a single LC-MS/MS run to combine DDA-based peptide identification with DIA-based extraction using deep learning–predicted spectral libraries for bottom-up proteomic analysis.

Key Features:

• Integration of DDA and DIA: Combines conventional DDA peptide identification with DIA scans in a single LC-MS/MS run to leverage complementary information from both acquisition modes.
• Deep learning-based spectral library generation: Uses deep learning LC-MS/MS property prediction tools to generate spectral libraries for DIA extraction.
• Reusable spectral libraries: Generates spectral libraries that can be repeatedly used to extract DIA scans for peptide identification and quantification.
• iRT vs RT calibration curve generation: Implements an iRT versus RT calibration module to align predicted and observed retention times.
• DIA extraction classifier training: Trains a classifier to improve specificity and sensitivity when extracting peptide signals from DIA data.
• False Discovery Rate (FDR) control: Applies FDR control mechanisms within the pipeline to limit false-positive identifications.
• Minimal information requirement: Operates with minimal required input information for data processing.

Scientific Applications:

• Enhanced proteomic profiling: Provides more comprehensive protein identification and quantification by combining DDA identifications with DIA extraction.
• Improved data quality and reliability: Leverages deep learning–derived spectral libraries, calibrated retention times, classifier-based extraction, and FDR control to increase confidence in results.
• Versatility in research applications: Applies to biomarker discovery, disease pathology studies, and functional genomics requiring detailed bottom-up proteomic analysis.

Methodology:

Combines DDA and DIA acquisition on LC-MS/MS; uses deep learning LC-MS/MS property prediction to generate spectral libraries and repeatedly extract DIA scans; performs iRT versus RT calibration curve generation; trains a DIA extraction classifier; and applies FDR control for identifications.