MS/MS Spectrum Prediction for Modified Peptides Using pDeep2 Trained by Transfer Learning

Abstract

In the past decade, tandem mass spectrometry (MS/MS)-based bottom-up proteomics has become the method of choice for analyzing post-translational modifications (PTMs) in complex mixtures. The key to the identification of the PTM-containing peptides and localization of the PTM-modified residues is to measure the similarities between the theoretical spectra and the experimental ones. An accurate prediction of the theoretical MS/MS spectra of the modified peptides will improve the similarity measurement. Here, we proposed the deep-learning-based pDeep2 model for PTMs. We used the transfer learning technique to train pDeep2, facilitating the training with a limited scale of benchmark PTM data. Using the public synthetic PTM data sets, including the synthetic phosphopeptides and 21 synthetic PTMs from ProteomeTools, we showed that the model trained by transfer learning was accurate (>80% Pearson correlation coefficients were higher than 0.9), and was significantly better than the models trained without transfer learning. We also showed that accurate prediction of the fragment ion intensities of the PTM neutral loss, for example, the phosphoric acid loss (−98 Da) of the phosphopeptide, will improve the discriminating power to distinguish the true phosphorylated residue from its adjacent candidate sites. pDeep2 is available at .

Data‐Independent Acquisition Mass Spectrometry‐Based Proteomics and Software Tools: A Glimpse in 2020

“…DeepMass [46] https://github.com/verilylifesciences/deepmass/tree/master/prism Prosit [45] https://www.proteomicsdb.org/prosit/ pDeep [ 47,84] https://github.com/pFindStudio/pDeep/tree/master/pDeep2…”

“…DeepRT [84] https://github.com/horsepurve/DeepRTplus MS GF+ [32] https://msgfplus.github.io/msgfplus/ OMSSA [86] https://sivome.github.io/proteomics/2019/03/02/Proteomics-with-OMSSA.html…”

“…MaxQuant [ 30,87] http://www.coxdocs.org/doku.php?id=maxquant:common:download_and_installation pFind [ 34,84] http://pfind.ict.ac.cn/software/pFind3/index.html…”

DeepRescore: Leveraging Deep Learning to Improve Peptide Identification in Immunopeptidomics

“…[13] The performance of Percolator is tightly associated with the features used for semi-supervised learning. Recent advancements in deep learning have enabled accurate peptide retention time (RT) prediction [15][16][17] and MS/MS spectrum prediction [15,18,19] for a given peptide sequence. We reasoned that adding these new, peptide-specific features to Percolator could further improve the confidence and sensitivity of peptide identification in immunopeptidomics data analysis.…”

DIAFree enables untargeted open-search identification for Data-Independent Acquisition data

“…In some studies, machine learning methods, such as pDeep (12,13) and Prosit (14,15), is employed to lessen those limitations and generate spectral libraries (7,14,16). Still, the presence of posttranslational modification (PTM) generates many peptidoforms (17)(18)(19), which makes the predicting process time-consuming.…”