DeepMS2-phospho

Using deep learning to generate in silico spectral libraries for site localization of phosphopeptides.

System Requirements

• Python >= 3.5. Anaconda is recommended.
• Keras with TensorFlow backend.
• R. As an alternative, the latest version of Microsoft R Open should be fine.
• RStudio is recommended but optional.

DeepMS2-phospho has been tested on a workstation with Intel Xeon E5-2690 v3 CPU, 16 GB RAM, and Microsoft Windows Server 2016 Version 1607 (OS Build 14393.2430) 64-bit operating system.

Installation

1. Install Python (Anaconda)

Download Anaconda Installer form https://www.anaconda.com/distribution/.

DeepMS2-phospho has been tested using Anaconda 4.2.0 (Python 3.5.2).

2. Install TensorFlow and Keras

Install TensorFlow using pip:

pip install --upgrade tensorflow

For GPU-supported version,

pip install --upgrade tensorflow-gpu

TensorFlow documentation is available at https://www.tensorflow.org/.

Install Keras:

pip install keras

Keras documentation is available at https://keras.io/.

DeepMS2-phospho has been tested using Keras 2.2.4 and TensorFlow 1.11.

3. Install R and RStudio

R is available at https://www.r-project.org/. As an alternative, Microsoft R Open is available at https://mran.microsoft.com/open/.

RStudio is available at https://www.rstudio.com/.

DeepMS2-phospho has been tested using Microsoft R Open 3.5.1 and RStudio 1.1.447.

Start R, ensure packages readr and rjson have been installed.

install.packages("readr")
install.packages("rjson")

Input Data Formats

1. Prepare a PSM List

A peptide-spectrum match (PSM) list is stored in a comma-separated values (CSV) file including the following columns: file, scan, charge and peptide.

An example file can be found at data/example/130418_03_SynthMix1.psm.csv.

"file","scan","charge","peptide"
"130418_03_SynthMix1_1pmol_10MS2_20min_NoDynExc",1708,2,"GTPSQS*PVVGR"
"130418_03_SynthMix1_1pmol_10MS2_20min_NoDynExc",1709,2,"ESLKEEDES*DDDNM@"
"130418_03_SynthMix1_1pmol_10MS2_20min_NoDynExc",1720,2,"ESLKEEDES*DDDNM@"
"130418_03_SynthMix1_1pmol_10MS2_20min_NoDynExc",1727,2,"GTPSQS*PVVGR"
"130418_03_SynthMix1_1pmol_10MS2_20min_NoDynExc",1734,2,"ESLKEEDES*DDDNM@"
...

In the peptide column, * represents phosphorylation (STY), and @ represents oxidation (M). Other modifications are not supported.

Users can generate PSM list files from database search tools or write them using Excel manually.

2. Prepare MS/MS Spectra

DeepMS2-phospho supports Mascot Generic Format (MGF) files as MS/MS spectra format. The file names of MGF files should match with those in the file column of the PSM list file (without extension).

Users can convert raw LC-MS/MS data to MGF files using MSConvert from ProteoWizard (https://github.com/ProteoWizard/pwiz).

An example MGF file can be found at data/example/130418_03_SynthMix1_1pmol_10MS2_20min_NoDynExc.mgf.

Spectral Library Generation

1. Prepare Peptide Lists

A peptide list is stored in a comma-separated values (CSV) file including the following columns: sequence and modification.

"sequence","modification"
"GTPSQSPVVGR","S4(Phospho)"
"GTPSQSPVVGR","S6(Phospho)"
"ESLKEEDESDDDNM","S2(Phospho)"
"ESLKEEDESDDDNM","S9(Phospho)"
"SPSPYYSR","S1(Phospho)"

Peptide list files can be generated from a PSM list file.

Start R and run init.R to load the functions.

source("{PATH_TO_CODE}/init.R")

Set the PSM list directory as working directory and run scripts/generate_phosphopeptides.R.

setwd("{PATH_TO_DATA}")
source("{PATH_TO_CODE}/scripts/generate_phosphopeptides.R")

Three peptide list files are output in the same directory.

• 130418_03_SynthMix1_Sphospho1.peptide.csv
• 130418_03_SynthMix1_Tphospho1.peptide.csv
• 130418_03_SynthMix1_Yphospho1.peptide.csv

2. Predict MS/MS Peak Intensities

Use DeepDIA to predicted MS/MS peak intensities from peptide lists.

For detailed manual, see https://github.com/lmsac/DeepDIA.

DeepDIA outputs the predicted JSON files.

• 130418_03_SynthMix1_Sphospho1_charge2.prediction.ions.json
• 130418_03_SynthMix1_Sphospho1_charge3.prediction.ions.json
• 130418_03_SynthMix1_Tphospho1_charge2.prediction.ions.json
• 130418_03_SynthMix1_Tphospho1_charge3.prediction.ions.json
• 130418_03_SynthMix1_Yphospho1_charge2.prediction.ions.json
• 130418_03_SynthMix1_Yphospho1_charge3.prediction.ions.json

User can also use MS2PIP, Prosit and pDeep2 for MS/MS peak intensity prediction. See docs for detailed manuals.

3. Generate Spectral Libraries

Set the prediction directory as working directory and run one of the following scripts:

• scripts/ions_to_msp.R: Convert the JSON files to spectral library files without any optimization.
• scripts/ions_to_msp_lossST.R: Convert the JSON files to spectral library files, in which phosphoric acid neutral loss peaks are generated for phosphorylated S/T.
• scripts/ions_to_msp_budding.R: Convert the JSON files to spectral library files with "budding" peaks.
• scripts/ions_to_msp_lossST_budding.R: Convert the JSON files to spectral library files with neutral loss and "budding" peaks.

setwd("{PATH_TO_DATA}")
source("{PATH_TO_CODE}/scripts/ions_to_msp[_lossST][_budding].R")

Spectral library (MSP) files are output in the same directory.

• 130418_03_SynthMix1_Sphospho1_charge2.prediction[_loss50][_bud100byp2].msp
• 130418_03_SynthMix1_Sphospho1_charge3.prediction[_loss50][_bud100byp2].msp
• 130418_03_SynthMix1_Tphospho1_charge2.prediction[_loss50][_bud100byp2].msp
• 130418_03_SynthMix1_Tphospho1_charge3.prediction[_loss50][_bud100byp2].msp
• 130418_03_SynthMix1_Yphospho1_charge2.prediction[_bud100by2].msp
• 130418_03_SynthMix1_Yphospho1_charge3.prediction[_bud100by2].msp

Phosphorylation Site Localization

1. Perform Spectral Matching

Open scripts/spectral_matching.R and change the parameters:

psmFile = '{PATH_TO_PSM_FILE}' # Path to the PSM list file
referencePath = '{PATH_TO_MSP}' # Path to the spectral library (MSP) directory
spectraPath = '{PATH_TO_MGF}' # Path to the MS/MS spectra (MGF) directory
resultFile = 'spectral_matching.result.csv' # Output file name

fragmentTolerance = 50 # m/z tolerance

Save and run the script.

source("{PATH_TO_CODE}/scripts/spectral_matching.R")

The spectral matching result file is output in the same directory.

• spectral_matching.result.csv

2. Output Result File

Place the PSM file and spectral matching result file in the working directory.

Run scripts/write_result.R.

source("{PATH_TO_CODE}/scripts/write_result.R")

Combining Spectral Matching with phosphoRS Results

Spectral matching is a sensitive method that supplements the probability-based approaches (e.g. phosphoRS) for phosphorylation site identification. Users can combine the spectral matching results with phosphoRS results. See docs for detailed manuals.

Publications

Yang, Y., Horvatovich, P., Qiao, L. Fragment mass spectrum prediction facilitates site localization of phosphorylation. J Proteome Res 20, 634–644 (2021). https://doi.org/10.1021/acs.jproteome.0c00580

License

DeepMS2-phospho functions (in code/functions) are distributed under a BSD license. See the LICENSE file for details. Scripts (in code/scripts) that require rjson and readr package are under a GPL-2 licence.

The example data (in data/example) are derived from a data set at ProteomeXchange (http://proteomecentral.proteomexchange.org/) with the accession key PXD000474. (Suni, V. et al. J. Proteome Res. 2015, 14 (5), 2348-2359.)

Contacts

Please report any problems directly to the github issue tracker. Also, you can send feedback to liang_qiao@fudan.edu.cn.