ProbMetab

Version: 1.0

Website

http://labpib.fmrp.usp.br/methods/probmetab/

Description

This package performs compound assignment on already peak grouped LC-MS data using bayesian modeling. It suggests using a combination of XCMS, CAMERA and mzMatch for data preprocessing. It provides the functionality to combine molecular ion by CAMERA annotation between acquision modes. A non-redundant set of molecular ion annotations can then be extracted. Included in the package is a database of compounds to reactions from KEGG (there is also the option a user importing their own compound to reaction database). If an assement of carbon offset is available this can be used to improve predictions. The likelihood of each mass to compound assignment is then calculated, using carbon isotope ratio information if available. The reactions between the identified metabolites are then codified. Analysis flux can also be incorporated at this stage. A reaction matrix can then be created from the annotated molecular ions and the database. The ratio between observed and theoretical isotope patterns is then calculated (if carbon labelling is available). The likelihood of each mass to compound assignment is then calculated using mass accuracy and isotopic pattern if available. The posterior probabilities are then calculated according to the model (using the Gibbs sampler). At this stage a table of identified metabolites can be exported as an R matrix. Identified metabolites are then mapped to pathways that can be visualised in cytoscape.

Functionality

• Annotation/MS/Level 3 - Tentative Candidates

Instrument Data Type

• MS/LC-MS

Approaches

Computer Skills

Advanced

Software Type

R Package

Interface

Command line interface

Operating System (OS)

• Unix/Linux
• Mac OS
• Windows

Language

R

Dependencies

CAMERA, graph, RCytoscape, multtest, RcppAramadillo, hwriter, GeneNet, RCurl, XML, rjson

Input Formats - Open

mzData, mzML, mzXML, netCDF

Input Formats - Proprietary

N/A

Published

2013

Last Updated

2013

License

GPL

Paper

http://www.ncbi.nlm.nih.gov/pubmed/24443383

PMID

24443383