biognosys logo

last modified 2017.06.27

mQuest/mProphet

Probabilistic scoring of MRM/SRM data

Multiple/Selected reaction monitoring is a technique for hypothesis driven proteomics research. Two major computational steps are involved when performing an MRM experiment:
  1. designing a sensitive and specific MRM assay
  2. processing the MRM measurements
mQuest/mProphet1 automates the second step, the analysis of MRM data, and provides probabilistic scoring of targeted peptide identifications and comprehensive quantification. mQuest/mProphet relies on the use of decoy transitions in the measurements which act like negative controls.

mQuest/mProphet is licensed under the Apache License Version 2.0. This software and associated documentation is provided "as is" and there is no warranty for this software.

The mProphet algorithm has been implemented into various software packages for the analysis of multiple reaction monitoring (MRM), parallel reaction monitoring (PRM) and data independent acquisition (DIA) or SWATH data. These software packages include SpectroDive (MRM/PRM), Spectronaut (DIA), OpenSWATH and Skyline.

1) Reiter, Lukas, et al. "mProphet: automated data processing and statistical validation for large-scale SRM experiments." Nature methods 8.5 (2011): 430-435.


Table of contents

Overview
Features
Data visualization
Software modules
Software installation
Software update
Releases
Howto analyze data
Manual and downloads
Authors

Overview

mQuest/mProphet is written in Perl and the R statistical programming language. mInteract coordinates the modules and will perform a data analysis with one command. Following input is necessary:
  • mzXML 
  • Transition list
A detailed description how to convert the raw data of Thermo and AB Sciex instruments to mzXML can be found in the manual (download section).
For the probabilistic part of the software some decoy transitions have to be measured in the experiment. Please, see the manual in the download section for more details.

The main output of the software is a spreadsheet containing all informtion such as transition meta data, confidence of identification, a transition interferrence score and quantification information.
The raw data and data analysis can also be browsed with a web browser.

Features

  • Automated processing of MRM data
  • Quality control (false discovery rate)
  • Quantification (height at apex and area)
  • Transition interference detection
  • Visualization
  • Browsing of raw data and analysis
  • Local installation
    or client server installation
  • Supports a large variety of workflows including:
    • label free
    • full proteome labelling
    • reference peptide spike in
    • "inverted" reference peptide spike in
  • Removal of transitions via web interface (e.g. because of interferences)
 

Data visualization

The software allows to browse the raw data and analysis with a web browser. Because SVG vector graphics is used to graphically display the analysis, Chrome or Opera is required as a web browser. The transitions are displayed including the meta data, scoring of the peak groups and the quality assignment.

Screenshot


screenshot

Software modules

mGen
generates decoy transitions for a list of target transitions. The data derived from the decoy transitions is used by mProphet to derive a statistical confidence of identification for the MRM signals.

mMap
links together the raw data (mzXML) with the MRM meta data (transition list table).

mQuest
scores the MRM data depending on the chosen workflow, performs the quantification and interference detection.

mProphet
combines the mQuest scores for optimal identification of MRM signals and derives a confidence of identificataion.

mInteract
coordinates mMap, mQuest and mProphet and allows a one command analysis.

Software installation

Virtual machine

We provide a virtual machine with the software already installed. The VM is configured to run locally (e.g. on a laptop). Prerequisites is a x64 CPU that supports virtualization (most modern CPUs fulfill this criterion). To install the VM please follow these steps:
  1. install VMware on your computer
  2. download the virtual machine
  3. unzip the virtual machine on your computer
  4. choose "Open a Virtual Machine" in the VMware Player Menu and browse to the unzipped folder
  5. choose "copied it" while importing the virtual machine
  6. run the virtual machine (the user name is 'mprophet', the password is 'mprophet')
You might have to turn on virtualization in your BIOS to get the virtual machine to run. See the VMware help for how to do this.

choose "example data analysis" on the desktop and click an .shtml file to browse through an example data set.

Visualization is enabled in Apache for the folder /usr/local/apps/biognosys/mquest/www. You can link any folder in there to enable visualization. This was already done for /home/mprophet/mProphet/mProphet_data_analysis.

To connect the virtual machine to the host file system and analyze experimental data on the virtual machine choose:
Virtual Machine -> Virtual Machine Settings -> Options -> Shared Folders
The folder will then be mounted in /mnt

Ubuntu Linux package

Installing the package with apt-get from a package repository will automatically install all the dependencies needed for the software. As a prerequisite you need a freely available 64bit Ubuntu Linux 10.04 installed. If no such operating system is present, Ubuntu can be installed into a virtual machine which makes a tedious dual boot system unnecessary. As a virtualization software one can for instance use the freely available VMware or VirtualBox.

There are two package types to choose for installation. Type one installs only the command line tools, type two includes a configured Apache2 server to be able to browse through the results with a web browser.

Please follow these steps for the installation:
  1. open a terminal on Ubuntu Linux
  2. append the following line to /etc/apt/sources.list:
    sudo gedit /etc/apt/sources.list
    line:
    deb http://systemsx03.ethz.ch/SybitDebRepo/binary/ubuntu10.04/DEBS/ /
    sudo will give you super user rights which is needed to edit the system file sources.list
  3. Generate the text file packager.key with the content from the link
    gedit packager.key 
  4. for security reasons it's obligatory to install a public key:
    sudo apt-key add packager.key
  5. update the repositories with the command:
    sudo apt-get update
  6. install either of the two package versions using the command:
    • sudo apt-get install mquestweb: recommended, Apache2 allows browsing through results
    • sudo apt-get install mquest: no Apache2 installation and hence no browsing through results
    for the web installation make sure to enter the correct domain name during installation (e.g. mquest.uni.edu or 131.63.230.61). If no host is entered when prompted, the visualization in the browser will be accessible locally via http://localhost. The web config file that specifies the host will be located in /usr/local/apps/biognosys/mquest/conf after installation.
Please be aware that the installation of mquestweb will modify your Apache configuration (httpd.conf,...)!

After installation the program will be located per default in /usr/local/apps/biognosys/mquest/bin. To be able to use shorter commands you can edit .profile with gedit ~/.profile and add the following line to the end of the file:
PATH=/usr/local/apps/biognosys/mquest/bin:$PATH
Activate the new profile:
source .profile


The standard path for data analysis is /usr/local/apps/biognosys/mquest/www. You can link any folder to this directory to be able to visualize the data e.g. with:
ln -s /home/<user>/mProhpet /usr/local/apps/biognosys/mquest/www
The path for this analysis in the web browser is http://localhost/mQuest-web/mProphet for a local installation.


We'd like to specially acknowledge Adam Srebniak for general help and making the package!
adam.srebniak (at) systemsx (pt) ch


Manual installation (Windows)

We recommend the installation using the package repository or to download the virtual machine (see above). However, the software can also be installed manually on Windows, Linux or Mac OS X. This, implies manual installation of dependencies (mainly perl modules). For data visualization of the raw data, a webserver (e.g. Apache) must also be installed. In the following we will not describe how to setup a webserver on windows. Here is a quick description on how to setup the command line tools on a windows machine:
  1. download the mQuest/mProphet software package from the download section
  2. install a perl interpreter (e.g. from http://www.activestate.com/activeperl)
  3. this is a list of non standard perl modules that are used by the software. they can be installed using the perl package manager ppm. Depending on the system, some modules might already be installed. Here is a screenshot with the typical modules that need to be installed on windows.
  4. install the R interpreter (http://cran.r-project.org/)
  5. start the R console and type install.packages("MASS") on the R command line
    Select a mirror and the R package will be installed
After this step all programs will properly run from the command line. To run the programs, change to the mQuest/bin directory and type perl mInteract.pl -help for a quick help. You can also add the path, where the scripts are, to the environment variable and run the programs from the command line without specifying the full path. For further information on how to use the software see the manual in the download section.

For data visualization a server that supports server side includes (SSI) must be installed. Although, in principle an Apache server can be installed and configured on windows we will not explain this here. Unless you are experienced with setting up a server we suggest to either install the package on an Ubuntu Linux or to install the virtual machine with the installed software.


Software update with Linux package

If there is a new release of the software it can updated on Ubuntu Linux as follows:
  1. sudo apt-get update
  2. sudo apt-get install mquest
    sudo apt-get install mquestweb
For how to have access to the package repository follow the steps 1-5 under Ubuntu Linux package above.

Following updating scenarios are possible:
  • mquest -> mquest
  • mquest -> mquestweb
  • mquestweb -> mquest
The web configuration can currently only be changed by hand or by complete removal of Apache and reinstallation of mquestweb.

To install a specific version of the package type e.g.:
sudo apt-get install mquest=V1.0.1.4

Package releases


name date
version repository revision changes
Sea Horse 23.12.2010 1.0.4.1 V1.0.4.1
848 - iRT support
Frog Fish 2.11.2011 2.0.3 V2.0.3 - auto detection of Biognosys RT-kit peptides
- graphical display of iRT
- removal of transition(s) on web interface
- mProphet d_score and m_score on web interface
- synthetic decoys
- auto detection of transition lists
- auto detection of newline type

Virtual machine releases


name date
version package version changes
Kukenan 4.10.2012 2.0 V1.0.4.1 packages installed

Howto analyze data

After successful installation of the package with apt-get on an Ubuntu Linux, the software is installed per default in the path:
/usr/local/apps/biognosys/mquest/bin/
In the following we describe howto analyze a data set derived from human plasma and reference peptides spiked into the sample. The data in the form of mrml.xml files can be downloaded below. Please see the manual for a description howto convert your mass spectometric data to mrml.xml.
  1. Add the above path to your path environment variable in your profile (e.g. ~/.profile) if you wish to directly call the programs without the path.
  2. The Apache server is configured such that results should be located in /usr/local/apps/biognosys/mquest/www/ in order to be able to browser through the results with a web browser. So, either make a directory in this path or link your working directory there e.g. with:
    ln -s /home/<user>/mprophet/ /usr/local/apps/biognosys/mquest/www/
  3. As an example data set download this experiment derived from human plasma samples and copy all data to /home/<user>/mprophet/
  4. Change to this directory and process the data with the following command:
    mInteract.pl -mquest -mprophet -workflow SPIKE_IN -mprophet_num_xval 20
    This will first run a peak detection. After that the peaks will be grouped into peak groups and these peak groups are then scored according to a number of criteria with mQuest. The resulting peak groups are then analyzed with mProphet. mProphet determines an optimal combination of scores to separate high quality from low quality signals. Finally a qvalue is added to the peak groups which allows filtering according to a false discovery rate. mProphet graphics is printed to mProphet.pdf, peak groups are written to mProphet_all_peakgroups.xls including quantification and the error statistics is written to mProphet_stat.xls.
  5. Go to http://localhost/mQuest-web/mprophet/ (this depends on your setup) with your web browser (Opera or Chrome are required). Click on the .shtml file to browse through the data.

Manual and downloads

  • detailed description of the software. How to convert raw data to mzXML, the transition list file format and more in the manual
  • decoy transition generator mGen as a standalone program
  • example input and output file of mGen
  • skyline transition export report format mProphet.skyr to use as input for mGen
  • mQuest/mProphet software for manual installation including mGen, mMap, mQuest and mProphet 
    dependencies have to be manually installed
  • mrml files of an experiment derived from human plasma

Authors



Lukas Reiter

current address

  Biognosys AG
  Wagistrasse 21
  CH-8952 Schlieren

former address
  Hengartner Laboratory
  Institute of Molecular Life Sciences
  Winterthurerstrasse 190
  University of Zurich - Irchel
  CH-8057 Zurich
former located
  Aebersold Laboratory
  Institute for Molecular Systems Biology
  Wolfgang-Pauli-Str. 16
  ETH Hönggerberg, HPT C 75
  CH-8093 Zurich
Oliver Rinner

current address

  Biognosys AG
  Wagistrasse 21
  CH-8952 Schlieren

former address
  Aebersold Laboratory
  Institute for Molecular Systems Biology
  Wolfgang-Pauli-Str. 16
  ETH Hönggerberg
  CH-8093 Zurich