Unleash the power of Hybrid-DIA data analysis with AI-driven software for biomarker discovery and validation in translational research

Introduction

Hybrid-DIA is a powerful mass spectrometry method that intelligently combines the strengths of Data Independent Acquisition (DIA) and Parallel Reaction Monitor (PRM) for biomarker discovery and validation in one single experiment [1,2]. It allows for unbiased proteome profiling, as well as targeted monitoring/quantifying of endogenous biomarkers using spiked-in counterparts, making it particularly valuable in translational research.

However, there is currently a lack of standardized software for processing Hybrid-DIA data. Existing publications require users to navigate through multiple software platforms to obtain DIA results, extract PRM signals, and perform signal normalization[1,2]. This fragmented approach is time-consuming and prone to errors, highlighting the need for a unified and automated software tool.

To address this, we propose an AI-driven software tool that simplifies Hybrid-DIA data analysis. Our tool integrates AI algorithms for DIA analysis, PRM signal extraction, and normalization, streamlining the process. Additionally, it utilizes DIA-de novo technology for novel peptide identification. Our software tool enhances efficiency and accuracy in Hybrid-DIA data analysis, unlocking its full potential for biomarker discovery and validation in translational research.

Methods

We used published dataset from MSVXXXX to benchmark our proposed workflow.

Preliminary Data or Plenary Speakers Abstract

For fitting Hybrid-DIA, we used AI-driven predictors and decision tree to select suitable fragment ions for PRM and DIA quantification, this increased sensitivity and reduce intense labour work.

First, we evaluated our workflow using simulated samples, which consist of human Hela cells spiked with isotope-labeled peptides representing 185 tumor-associated antigens (TAA). The results, depicted in Figure X, demonstrate that Hybrid-DIA outperformed conventional DIA in identifying and quantifying TAAs. Hybrid-DIA successfully identified and quantified 167 out of the 185 TAAs, especially excelling in low input samples where DIA failed to extract signals, highlighting the comparable sensitivity of Hybrid-DIA to Parallel Reaction Monitor (PRM).

Furthermore, real-world data from patient samples was also analyzed. Our workflow identified an average of 6500 protein groups in both DIA and Hybrid-DIA data. Principal Component Analysis (PCA) plots and protein heatmaps revealed significant differences across different patients.

Finally, DIA de novo also reported some interesting peptides which is not listed in the fasta. We check one peptide which may derived from mutation and validate its sequence by synthetic peptide.

In summary, our streamlined Hybrid-DIA workflow offers enhanced sensitivity and performance for biomarker discovery and quantitation analysis, particularly in low input samples. Additionally, DIA de novo helps to find potential mutations whose sequence is not recorded in protein database.