patRoon aims to provide a solution for comprehensive mass spectrometry based non-target analysis (NTA) workflows for environmental analysis. The name is derived from a Dutch word that means pattern and may also be an acronym for hyPhenated mAss specTROmetry nOn-target aNalysis.
Mass spectrometry based non-target analysis is used to screen large numbers of chemicals simultaneously. For this purpose, high resolution mass spectrometry instruments are used which are typically coupled (or hyphenated) with chromatography (e.g. LC or GC) to provide additional separation of analytes prior to mass detection. The size and complexity of resulting data makes manual processing impractical. Many dedicated software tools were developed and are still being developed to facilitate a more automated approach. However, these tools are generally not optimized for environmental workflows and/or only implement parts of the functionality required for a complete NTA workflow. An important aim of patRoon is to harmonize the many tools available in a consistent user interface, which removes the need to know all the details of each individual software tool and performing tedious conversion of data when combining tools in a workflow. In addition, key functionality is implemented in patRoon to provide complete NTA workflows, such as fully automated annotation, evaluating and combining of results from different algorithms, automatic reporting, several user friendly graphical tools, extensive documentation, several important optimization strategies and more. Many of the available tools nowadays were developed using the R language or are otherwise easily interface with R, hence, it was a straightforward decision to develop patRoon as an R package.
The workflow of non-target analysis is typically highly dependent on several factors such as the analytical instrumentation used and requirements of the study. For this reason, patRoon does not enforce a certain workflow. Instead, most workflow steps are optional, are highly configurable and algorithms can easily be mixed or even combined.
Below is an overview of the most common non-target workflow steps supported along with various software tools that were used to implement them:
- Data preparation: conversion between and export to common open MS data formats (e.g. mzXML and mzML) (ProteoWizard, OpenMS, DataAnalysis).
- Extraction and grouping of features (e.g. XCMS, OpenMS, enviPick, ProfileAnalysis).
- Data cleanup: post filtering of data to improve its quality and several tools to aid prioritization.
- Automatic extraction of MS and MS/MS data (mzR, DataAnalysis).
- Formula calculation: automatic calculation of candidate formulae for detected features (GenForm, SIRIUS, DataAnalysis).
- Compound identification: automatic annotation of MS/MS spectra and retrieval of candidate structures (MetFrag, SIRIUS with CSI:FingerID).
- Generation of components: grouping of (chemically) related features such as isotopes, adducts and homologs (RAMClustR, CAMERA, nontarget R package) and with related intensity profiles.
- reporting of all workflow steps in various formats: CSV, PDF and HTML.
An example HTML report which gives an overview of what data can be produced can viewed here.
- Developed on both Windows and Linux
data.tableis used internally as a generally much more efficient alternative todata.frame.- The processx R package is used to execute command line processes in parallel to reduce computation times.
- Results from workflow steps are cached within a SQLite database to avoid repeated computations.
- The RDCOMClient is used to provide functionality from DataAnalysis (Bruker).
- Supports all major instrument vendor input formats (through usage of ProteoWizard and DataAnalysis).
- The Shiny R package was used to implement several GUI tools.
- S4 classes and generics are used to implement a consistent interface to the many different supported software algorithms for each workflow step.
- Continuous integration is used for automated unit testing, automatically updating the Website and documentation and maintaining a miniCRAN repository and Docker image to simplify installation (see the handbook for more details).
patRoon itself can be installed as any other R package, however, depending on which algorithms you want to use in your workflow, some extra steps may be required to install all the necessary tools. Please see the installation section in the handbook for more information.
For a very quick start:
library(patRoon)
newProject()The newProject() function will pop-up a dialog screen (requires R Studio!) which will allow you to quickly select the analyses and common workflow options to subsequently generate a template R processing script.
However, for a better guide to get started it is recommended to read the tutorial. Afterwards the handbook is a recommended read if you want to know more about advanced usage of patRoon. Finally, the reference outlines all the details of the patRoon package.