Updated MS²PIP web server supports cutting-edge proteomics applications

Abstract

Interest in the use of machine learning for peptide fragmentation spectrum prediction has been strongly on the rise over the past years, especially for applications in challenging proteomics identification workflows such as immunopeptidomics and the full-proteome identification of data independent acquisition spectra. Since its inception, the MS²PIP peptide spectrum predictor has been widely used for various downstream applications, mostly thanks to its accuracy, ease-of-use, and broad applicability. We here present a thoroughly updated version of the MS²PIP web server, which includes new and more performant prediction models for both tryptic- and non-tryptic peptides, for immunopeptides, and for CID-fragmented TMT-labeled peptides. Additionally, we have also added new functionality to greatly facilitate the generation of proteome-wide predicted spectral libraries, requiring only a FASTA protein file as input. These libraries also include retention time predictions from DeepLC. Moreover, we now provide pre-built and ready-to-download spectral libraries for various model organisms in multiple DIA-compatible spectral library formats. Besides upgrading the back-end models, the user experience on the MS²PIP web server is thus also greatly enhanced, extending its applicability to new domains, including immunopeptidomics and MS3-based TMT quantification experiments. MS²PIP is freely available at https://iomics.ugent.be/ms2pip/.

Thunder-DDA-PASEF enables high-coverage immunopeptidomics and is boosted by MS2Rescore with MS2PIP timsTOF fragmentation prediction model

“…Briefly, we enriched HLAIps from JY cells by immunoprecipitation using W6/32 antibody, and analyzed them by nanoLC-IMS-MS on a nanoElute coupled to timsTOF-Pro-2 in DDA-PASEF mode, using PEAKS XPro for subsequent peptide identification. After training an MS 2 PIP model 28 to predict peptide fragmentation for timsTOF data, we used it in an updated version of MS 2 Rescore 16 , 29 (v3.0.0b4) to improve the identifications for the dataset exploring the SARS-Cov-2 spike immunopeptidome. To evaluate the identification of possible HLA class I ligands, we predicted the peptide binding to the respective HLA alleles of each sample using NetMHCpan-4.1 30 via MhcVizPipe 31 .…”

“…d Data analysis: Database search was performed in PEAKS XPro using unspecific cleavage. After training a MS 2 PIP 28 timsTOF fragmentation prediction model, peptide identification was rescored using MS 2 Rescore (MS 2 R, v3.0.0b4) 16 , 29 . Data analysis was performed in R and predicted MHC-binding affinity was evaluated using NetMHCpan-4.1 30 and GibbsCluster-2.0 64 through MhcVizPipe (v0.7.9) 31 .…”

“…To improve the peptide identification, we implemented MS 2 Rescore (MS 2 R, v3.0.0b4) 16 , 29 into our workflow. Since peak intensity predictions rely heavily on the fragmentation method, instrument or labeling method 28 , 45 , we trained a timsTOF-specific peak intensity model using in-house timsTOF data. In addition, it has been shown that prediction models trained on datasets including both, tryptic and non-tryptic peptides, can perform equally well for predicting the fragmentation of both types of peptides 16 , 34 .…”

“…In addition, it has been shown that prediction models trained on datasets including both, tryptic and non-tryptic peptides, can perform equally well for predicting the fragmentation of both types of peptides 16 , 34 . Therefore, we trained a MS 2 PIP 28 timsTOF fragmentation prediction model using data acquired at two different labs from immunopeptidomics (HLA class I), tryptic peptides, and elastase digest samples. In total, 241,104 peptides (including modifications) were used to train the model and a distinct set of 10,045 peptides to test it.…”

“…Compared to the Standard method, Thunder-DDA-PASEF doubled (on average) the HLAIps identifications across samples with diverse HLA alleles. In addition, to increase the number and confidence of peptides identified, we trained a specialized MS 2 PIP model 28 to predict peptide fragmentation in timsTOF instruments. Using this model to rescore the results with MS 2 Rescore (v3) 16 , 29 boosted HLAIp identifications by 41.7% to 33%, resulting in 5738 HLAIps detected from as little as one million JY cell equivalents, and 14,516 HLAIps from 20 million.…”