IPF

Overview

IPF (Inference of PeptidoForms) [1] is an extension to the OpenSWATH [2] workflow to increase the specificity of the analysis to the level of peptidoforms (modified peptides with specific site-localization) across multiple runs. IPF is fully implemented as part of OpenMS [3] and PyProphet [4] and compatible with the downstream alignment algorithm TRIC [5].

Contact and Support

We provide support for IPF using the OpenMS support channels. Please address general questions to the open-ms-general mailing list.

You can contact the author George Rosenberger.

Installation

IPF requires the installation of several tools:

OpenMS

IPF is available since OpenMS 2.1. Please follow the instructions in the OpenSWATH tutorial to install OpenMS.

Percolator

IPF presently requires QVALITY [6] to be installed, which is part of Percolator. Please install and export the binaries according to the Percolator instructions.

PyProphet

IPF requires a specific branch of PyProphet. If Python and PIP are configured correctly, the following command can be used to install the latest version:

pip install git+https://github.com/grosenberger/pyprophet.git@feature/ipf

To work properly, PyProphet needs to find the qvality executable in the main path.

TRIC

TRIC should be installed according to the TRIC installation instructions.

Tutorial

Running IPF requires following several steps:

1. Peptide Query Parameter Generation

IPF requries a spectral library generated from DDA (or DIA pseudo spectra, e.g. from DIA-Umpire [7]) data. The protocol is described in more detail in the Trans-Proteomic Pipeline tutorial.

Using SpectraST, the spectral library db_consensus.splib is converted to a MRM and then to a TraML file:

# This will generate the file db_assays.mrm
spectrast -cNdb_assays -cICID-QTOF -cM db_consensus.splib

ConvertTSVToTraML -in db_assays.mrm -out db_assays.TraML

Then, a TraML file containing the detection and identification transitions is being generated. At this step, the residue modifiability needs to be defined in the OpenMS resource directory.

Danger

To specificy the residue modifiability, the files CHEMISTRY/PSI-MOD.obo and CHEMISTRY/unimod.xml in the OpenMS resource directory can be manually modified. This is a common source of problems downstream in the analysis, because the tools will not complain if the folder was not found or the files are incorrect, but will rather will use the default files.

If the residue modifiability for e.g. phosphorylation should be changed, make sure that both files are modified. An example for phosphorylation can be obtained from the ProteomeXchange repository. The location (IMPORTANT: MUST BE AN ABSOLUTE PATH!) of the modified OpenMS resource directory needs to be supplied by setting OPENMS_DATA_PATH for OpenSwathAssayGenerator.

OPENMS_DATA_PATH=/modified_path/share \
OpenSwathAssayGenerator -in db_assays.TraML \
-out db_assays_ptms.TraML \
-swath_windows_file swath64.txt \
-allowed_fragment_charges 1,2,3,4 \
-enable_ms1_uis_scoring \
-max_num_alternative_localizations 2000 \
-enable_identification_specific_losses \
-enable_identification_ms2_precursors

Make sure to double-check this step. If the OPENMS_DATA_PATH is not set correctly, NO error message or warning will appear, but OpenSwathAssayGenerator will fall back to the default settings. To ensure that everything worked correctly, consider running OpenSwathAssayGenerator twice, once with the original residue modifiability files and once with the modified ones. The resulting files need to be different.

We then append decoys to the library:

OPENMS_DATA_PATH=/modified_path/share \
OpenSwathDecoyGenerator -in db_assays_ptms.TraML \
-out db_assays_ptms_decoys.TraML \
-method shuffle \
-append \
-mz_threshold 0.1 \
-remove_unannotated

2. Targeted data extraction using OpenSWATH

The next step is conducted using OpenSWATH.

OPENMS_DATA_PATH=/modified_path/share \
OpenSwathWorkflow -min_upper_edge_dist 1 \
-mz_extraction_window 0.05 \
-rt_extraction_window 600 \
-extra_rt_extraction_window 100 \
-min_rsq 0.95 \
-min_coverage 0.6 \
-use_ms1_traces \
-enable_uis_scoring \
-Scoring:uis_threshold_peak_area 0 \
-Scoring:uis_threshold_sn 0 \
-Scoring:stop_report_after_feature 5 \
-tr_irt DIA_iRT.TraML \
-tr db_assays_ptms_decoys.TraML \
-threads 8 \
-in MSDATA.mzXML.gz \
-out_tsv MSDATA_RESULTS.tsv

Important is to set the parameters -use_ms1_traces and -enable_uis_scoring to extract the additional identification transitions and precursor signals using OpenSWATH.

3. Statistical validation using PyProphet

PyProphet is then applied to the OpenSWATH results:

pyprophet --target.overwrite \
--final_statistics.emp_p \
--qvality.enable \
--qvality.generalized \
--ms1_scoring.enable \
--uis_scoring.enable \
--d_score.cutoff=100000 \
--semi_supervised_learner.num_iter=20 \
--xeval.num_iter=20 \
--ignore.invalid_score_columns \
--uis_scoring.expand_peptidoforms MSDATA_RESULTS.tsv

It generates reports on several different levels. Important for TRIC are the files that end with *_uis_expanded.csv. IPF attaches several columns, e.g. PosteriorFullPeptideName, which contains the peptidoform sequence of the best scoring peptidoform. The column pfqm_score represents the peptidoform q-value, whereas pf_score represent the posterior probability. After running IPF, the m_score column is equal to pfqm_score to enable alignment by TRIC.

4. Multi-run alignment using TRIC

TRIC can be applied to the IPF results with the following command:

feature_alignment.py --in *_uis_expanded.csv \
--out feature_alignment.csv \
--out_matrix feature_alignment_matrix.csv \
--file_format openswath \
--fdr_cutoff 0.01 \
--max_fdr_quality 0.2 \
--mst:useRTCorrection True \
--mst:Stdev_multiplier 3.0 \
--method LocalMST \
--max_rt_diff 30 \
--alignment_score 0.0001 \
--frac_selected 0 \
--realign_method lowess_cython \
--disable_isotopic_grouping

Data

Availability

The synthetic phosphopeptide reference mass spectrometry proteomics data is available from PRIDE/ProteomeXchange with the data set identifier PXD004573.

The enriched U2OS phosphopeptide mass spectrometry proteomics data is available from PRIDE/ProteomeXchange with the data set identifier PXD006056.

The 14-3-3β phosphopeptide interactomics mass spectrometry proteomics data is available from PRIDE/ProteomeXchange with the data set identifier PXD006057.

The twin study mass spectrometry proteomics data is available from PRIDE/ProteomeXchange with the data set identifier PXD004574.

References

[1]Rosenberger G, Liu Y, Röst HL, Ludwig C, Buil A, Bensimon A, Soste M, Spector TD, Dermitzakis ET, Collins BC, Malmström L, Aebersold R. Inference and quantification of peptidoforms in large sample cohorts by SWATH-MS. Nat Biotechnol. 2017 Aug;35(8):781-788. doi: 10.1038/nbt.3908. Epub 2017 Jun 12. PMID: 28604659
[2]Röst HL, Rosenberger G, Navarro P, Gillet L, Miladinović SM, Schubert OT, Wolski W, Collins BC, Malmström J, Malmström L, Aebersold R. OpenSWATH enables automated, targeted analysis of data-independent acquisition MS data. Nat Biotechnol. 2014 Mar 10;32(3):219-23. doi: 10.1038/nbt.2841. PMID: 24727770
[3]Röst HL, Sachsenberg T, Aiche S, Bielow C, Weisser H, Aicheler F, Andreotti S, Ehrlich HC, Gutenbrunner P, Kenar E, Liang X, Nahnsen S, Nilse L, Pfeuffer J, Rosenberger G, Rurik M, Schmitt U, Veit J, Walzer M, Wojnar D, Wolski WE, Schilling O, Choudhary JS, Malmström L, Aebersold R, Reinert K, Kohlbacher O. OpenMS: a flexible open-source software platform for mass spectrometry data analysis. Nat Methods. 2016 Aug 30;13(9):741-8. doi: 10.1038/nmeth.3959. PMID: 27575624
[4]Teleman J, Röst HL, Rosenberger G, Schmitt U, Malmström L, Malmström J, Levander F. DIANA–algorithmic improvements for analysis of data-independent acquisition MS data. Bioinformatics. 2015 Feb 15;31(4):555-62. doi: 10.1093/bioinformatics/btu686. Epub 2014 Oct 27. PMID: 25348213
[5]Röst HL, Liu Y, D’Agostino G, Zanella M, Navarro P, Rosenberger G, Collins BC, Gillet L, Testa G, Malmström L, Aebersold R. TRIC: an automated alignment strategy for reproducible protein quantification in targeted proteomics. Nat Methods. 2016 Sep;13(9):777-83. doi: 10.1038/nmeth.3954. Epub 2016 Aug 1. PMID: 27479329
[6]Käll L, Storey JD, Noble WS. QVALITY: non-parametric estimation of q-values and posterior error probabilities. Bioinformatics. 2009 Apr 1;25(7):964-6. doi: 10.1093/bioinformatics/btp021. Epub 2009 Feb 4. PMID: 19193729
[7]Tsou CC, Avtonomov D, Larsen B, Tucholska M, Choi H, Gingras AC, Nesvizhskii AI. DIA-Umpire: comprehensive computational framework for data-independent acquisition proteomics. Nat Methods. 2015 Mar;12(3):258-64, 7 p following 264. doi: 10.1038/nmeth.3255. Epub 2015 Jan 19. PMID: 25599550