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

ABSTRACT: Here, a workflow of 4D DIA proteomics by using the predicted multi-dimensional in silico library was established. A deep learning model Deep4D that could high-accurately predict the CCS and RT of both the unmodified and phosphorylated peptides was developed, which is the model based on self-attention14 that completely avoid the use of recurrent neural network (RNN) or LSTM. Deep4D exhibited higher accuracy in the prediction of CCS and RT of peptides than the current models based on deep learning. By using Deep4D and MS/MS prediction tool, an integrated in silico library containing CCS, RT, and fragment ion intensities of millions of peptides for 4D DIA proteomics was established based on the SwissProt H. sapiens database, which enables the deeper peptide and proteome coverage for human samples compared to using sample-specific experimental library. We further demonstrate that the introduction of in silico prediction library can greatly complement the experimental library of directly obtained phosphorylated peptides, resulting in a greater increase in the identification of phosphorylated peptides and phosphorylated proteins.

ORGANISM(S): Homo Sapiens Mus Musculus Saccharomyces Cerevisiae

SUBMITTER: Suming Chen  

PROVIDER: PXD034553 | iProX | Wed Apr 19 00:00:00 BST 2023

REPOSITORIES: iProX