ms2pip 3.13.0

MS²PIP: MS² Peak Intensity Prediction - Fast and accurate peptide fragmention spectrum prediction for multiple fragmentation methods, instruments and labeling techniques.

• Introduction
• Installation
• Usage
• Specialized prediction models

Introduction

MS²PIP is a tool to predict MS² peak intensities from peptide sequences. The result is a predicted peptide fragmentation spectrum that accurately resembles its observed equivalent. These predictions can be used to validate peptide identifications, generate proteome-wide spectral libraries, or to select discriminative transitions for targeted proteomics. MS²PIP employs the XGBoost machine learning algorithm and is written in Python.

You can install MS²PIP on your machine by following the installation instructions below. For a more user-friendly experience, go to the MS²PIP web server. There, you can easily upload a list of peptide sequences, after which the corresponding predicted MS² spectra can be downloaded in multiple file formats. The web server can also be contacted through the RESTful API.

To generate a predicted spectral library starting from a FASTA file, we developed a pipeline called fasta2speclib. Usage of this pipeline is described on the fasta2speclib wiki page. Fasta2speclib was developed in collaboration with the ProGenTomics group for the MS²PIP for DIA project.

To improve the sensitivity of your peptide identification pipeline with MS²PIP predictions, check out MS²Rescore.

If you use MS²PIP for your research, please cite the following publication:

• Gabriels, R., Martens, L., & Degroeve, S. (2019). Updated MS²PIP web server delivers fast and accurate MS² peak intensity prediction for multiple fragmentation methods, instruments and labeling techniques. Nucleic Acids Research doi:10.1093/nar/gkz299

Prior MS²PIP publications:

• Degroeve, S., Maddelein, D., & Martens, L. (2015). MS²PIP prediction server: compute and visualize MS² peak intensity predictions for CID and HCD fragmentation. Nucleic Acids Research, 43(W1), W326–W330. doi:10.1093/nar/gkv542
• Degroeve, S., & Martens, L. (2013). MS²PIP: a tool for MS/MS peak intensity prediction. Bioinformatics (Oxford, England), 29(24), 3199–203. doi:10.1093/bioinformatics/btt544

Please also take note of, and mention, the MS²PIP version you used.

Installation

Pip package

With Python 3.6 or higher, run:

pip install ms2pip

Compiled wheels are available for Python 3.6, 3.7, and 3.8, on 64bit Linux, Windows, and macOS. This should install MS²PIP in a few seconds. For other platforms, MS²PIP can be built from source, although it can take a while to compile the large prediction models.

We recommend using a venv or conda virtual environment.

Conda package

Install with activated bioconda and conda-forge channels:

conda install -c defaults -c bioconda -c conda-forge ms2pip

Bioconda packages are only available for Linux and macOS.

Docker container

First check the latest version tag on biocontainers/ms2pip/tags. Then pull and run the container with

docker container run -v <working-directory>:/data -w /data quay.io/biocontainers/ms2pip:<tag> ms2pip <ms2pip-arguments>

where <working-directory> is the absolute path to the directory with your MS²PIP input files, <tag> is the container version tag, and <ms2pip-arguments> are the ms2pip command line options (see Command line interface).

For development

Clone this repository and use pip to install an editable version:

pip install --editable .

Usage

1. Fast prediction of large amounts of peptide spectra
   1. Command line interface
   2. Python API
   3. Input files
      1. Config file
      2. PEPREC file
      3. Spectrum file (optional)
      4. Examples
   4. Output
2. Predict and plot a single peptide spectrum

Fast prediction of large amounts of peptide spectra

MS²PIP comes with pre-trained models for a variety of fragmentation methods and modifications. These models can easily be applied by configuring MS²PIP in the config file and providing a list of peptides in the form of a PEPREC file. Optionally, MS²PIP predictions can be compared to observed spectra in an MGF or mzmL file.

Command line interface

To predict a large amount of peptide spectra, use ms2pip:

usage: ms2pip [-h] -c CONFIG_FILE [-s SPECTRUM_FILE] [-w FEATURE_VECTOR_OUTPUT]
       [-r] [-x] [-m] [-t] [-n NUM_CPU]
       [--sqldb-uri SQLDB_URI]
       <PEPREC file>

positional arguments:
  <PEPREC file>         list of peptides

optional arguments:
  -h, --help            show this help message and exit
  -c, --config-file     Configuration file: text-based (extensions `.txt`,
                        `.config`, or `.ms2pip`) or TOML (extension `.toml`).
  -s, --spectrum-file   MGF or mzML spectrum file (optional)
  -w, --vector-file     write feature vectors to FILE.{pkl,h5} (optional)
  -r, --retention-time  add retention time predictions (requires DeepLC python package)
  -x, --correlations    calculate correlations (if spectrum file is given)
  -m, --match-spectra   match peptides to spectra based on predicted spectra (if spectrum file is given)
  -n, --num-cpu         number of CPUs to use (default: all available)
  --sqldb-uri           use sql database of observed spectra instead of spectrum files
  --model-dir           custom directory for downloaded XGBoost model files. By default, `~/.ms2pip` is used.

Python API

The MS2PIP class can be imported from ms2pip.ms2pipC and run as follows:

>>> from ms2pip.ms2pipC import MS2PIP
>>> params = {
...     "ms2pip": {
...         "ptm": [
...             "Oxidation,15.994915,opt,M",
...             "Carbamidomethyl,57.021464,opt,C",
...             "Acetyl,42.010565,opt,N-term",
...         ],
...         "frag_method": "HCD",
...         "frag_error": 0.02,
...         "out": "csv",
...         "sptm": [], "gptm": [],
...     }
... }
>>> ms2pip = MS2PIP("test.peprec", params=params, return_results=True)
>>> predictions = ms2pip.run()

Input files

Config file

Several MS²PIP options need to be set in this config file.

• model=X where X is one of the currently supported MS²PIP models (see Specialized prediction models).
• frag_error=X where is X is the fragmentation spectrum mass tolerance in Da (only relevant if a spectrum file is passed).
• out=X where X is a comma-separated list of a selection of the currently supported output file formats: csv, mgf, msp, spectronaut, or bibliospec (SSL/MS2, also for Skyline). For example: out=csv,msp.
• ptm=X,Y,opt,Z for every peptide modification where:
  • X is the PTM name and needs to match the names that are used in the PEPREC file). If the --retention_time option is used, PTM names must match the PSI-MOD/Unimod names embedded in DeepLC (see DeepLC documentation).
  • Y is the mass shift in Da associated with the PTM.
  • Z is the one-letter code of the amino acid AA that is modified by the PTM. For N- and C-terminal modifications, Z should be N-term or C-term, respectively.

PEPREC file

To apply the pre-trained models you need to pass only a <PEPREC file> to MS²PIP. This file contains the peptide sequences for which you want to predict peak intensities. The file is space separated and contains at least the following four columns:

• spec_id: unique id (string) for the peptide/spectrum. This must match the TITLE field in the corresponding MGF file, or nativeID (MS:1000767) in the corresponding mzML file, if given.
• modifications: Amino acid modifications for the given peptide. Every modification is listed as location|name, separated by a pipe (|) between the location, the name, and other modifications. location is an integer counted starting at 1 for the first AA. 0 is reserved for N-terminal modifications, -1 for C-terminal modifications. name has to correspond to a modification listed in the Config file. Unmodified peptides are marked with a hyphen (-).
• peptide: the unmodified amino acid sequence.
• charge: precursor charge state as an integer (without +).

Peptides must be strictly longer than 2 and shorter than 100 amino acids and cannot contain the following amino acid one-letter codes: B, J, O, U, X or Z. Peptides not fulfilling these requirements will be filtered out and will not be reported in the output.

In the conversion_tools folder, we provide a host of Python scripts to convert common search engine output files to a PEPREC file.

To start from a FASTA file, see fasta2speclib.

Spectrum file (optional)

Optionally, an MGF or mzML file with measured spectra can be passed to MS²PIP. In this case, MS²PIP will calculate correlations between the measured and predicted peak intensities. Make sure that the PEPREC spec_id matches the MGF TITLE field or mzML nativeID. Spectra present in the spectrum file, but missing in the PEPREC file (and vice versa) will be skipped.

Examples

Suppose the config file contains the following lines

model=HCD
frag_error=0.02
out=csv,mgf,msp
ptm=Carbamidomethyl,57.02146,opt,C
ptm=Acetyl,42.010565,opt,N-term
ptm=Glyloss,-58.005479,opt,C-term

then the PEPREC file could look like this:

spec_id modifications peptide charge
peptide1 - ACDEK 2
peptide2 2|Carbamidomethyl ACDEFGR 3
peptide3 0|Acetyl|2|Carbamidomethyl ACDEFGHIK 2

In this example, peptide3 is N-terminally acetylated and carries a carbamidomethyl on its second amino acid.

The corresponding (optional) MGF file can contain the following spectrum:

BEGIN IONS
TITLE=peptide1
PEPMASS=283.11849750978325
CHARGE=2+
72.04434967 0.00419513
147.11276245 0.17418982
175.05354309 0.03652963
...
END IONS

Output

The predictions are saved in the output file(s) specified in the config file. Note that the normalization of intensities depends on the output file format. In the CSV file output, intensities are log2-transformed. To "unlog" the intensities, use the following formula: intensity = (2 ** log2_intensity) - 0.001.

Predict and plot a single peptide spectrum

With ms2pip-single-prediction a single peptide spectrum can be predicted with MS²PIP and plotted with spectrum_utils. For instance,

ms2pip-single-prediction "PGAQANPYSR" "-" 3 --model TMT

results in:

Run ms2pip-single-prediction --help for more details.

Specialized prediction models

MS²PIP contains multiple specialized prediction models, fit for peptide spectra with different properties. These properties include fragmentation method, instrument, labeling techniques and modifications. As all of these properties can influence fragmentation patterns, it is important to match the MS²PIP model to the properties of your experimental dataset.

Currently the following models are supported in MS²PIP: HCD, CID, iTRAQ, iTRAQphospho, TMT, TTOF5600, HCDch2 and CIDch2. The last two "ch2" models also include predictions for doubly charged fragment ions (b++ and y++), next to the predictions for singly charged b- and y-ions.

MS² acquisition information and peptide properties of the models' training datasets

Models, version numbers, and the train and test datasets used to create each model

To train custom MS²PIP models, please refer to Training new MS²PIP models on our Wiki pages.