High-Coverage Four-Dimensional Data-Independent Acquisition Proteomics and Phosphoproteomics Enabled by Deep Learning-Driven Multidimensional Predictions

Abstract

Four-dimensional (4D) data-independent acquisition (DIA)-based proteomics is a promising technology. However, its full performance is restricted by the time-consuming building and limited coverage of a project-specific experimental library. Herein, we developed a versatile multifunctional deep learning model Deep4D based on self-attention that could predict the collisional cross section, retention time, fragment ion intensity, and charge state with high accuracies for both the unmodified and phosphorylated peptides and thus established the complete workflows for high-coverage 4D DIA proteomics and phosphoproteomics based on multidimensional predictions. A 4D predicted library containing ∼2 million peptides was established that could realize experimental library-free DIA analysis, and 33% more proteins were identified than using an experimental library of single-shot measurement in the example of HeLa cells. These results show the great values of the convenient high-coverage 4D DIA proteomics methods.

Dithiocarbazate–Zn(ii) complexes for photodynamic therapy and chemotherapy against lung cancer

“…Four-dimensional data-independent acquisition (4D-DIA) proteomic analysis is a fast, efficient, and sensitive method that can help elucidate potential anticancer mechanisms of drugs at the protein level. 76,77 Therefore, we performed 4D-DIA proteomic analysis in the diaPASEF mode to understand the details of the chemotherapeutic mechanism of complex Zn3 .…”

Acquisition and Analysis of DIA-Based Proteomic Data: A Comprehensive Survey in 2023

“…These are now routinely applied to the identification of modified and unmodified peptides in DIA- and DDA-based proteomics ( 126 , 154 , 160 , 165 , 169 , 170 ). Meanwhile, many tools have been developed to streamline library prediction by integrating multiple steps ( 142 , 154 , 160 , 165 , 171 , 172 , 173 ). The topic about library generation through in silico prediction has been extensively overviewed in recent reviews ( 174 , 175 , 176 , 177 ).…”

3D-MPEA: A Graph Attention Model-Guided Computational Approach for Annotating Unknown Metabolites in Interactomics via Mass Spectrometry-Focused Multilayer Molecular Networking

“…The dynamic changes information is a high-dimensional readout of the changes in the biochemical reaction system that provides a signature about the biological processes affected by the perturbation-induced phenotypic changes. As a quality information collection strategy, which is extraordinarily well matched with multidimensional omics investigation, the emerging of variable data-independent acquisition (vDIA) enables the acquisition of enormous mass information through fragmenting all parent ions within a predetermined mass range. − However, due to the DIA-based omics generated copious highly multiplexed data and the mixed spectra produced by various precursor ions, manual curation of the results is hardly an option. MS/MS molecular network construction, on the basis of spectral similarity and homologous mass spectrometry fragments, is considered as an effective approach in qualification and mining of unknown compounds in complex matrices .…”