ProbMetab

ProbMetab annotates LC–MS metabolomics data using a Bayesian probabilistic framework to assign candidate metabolites to mass peaks and relate them via biochemical reaction networks.

Key Features:

• Bayesian Inference Engine: Implements a Bayesian model that integrates multiple sources of experimental data and metadata and supports flexible calculation of likelihood functions.
• Dirichlet-categorical Prior Distributions: Uses Dirichlet-categorical priors to enable selection of biologically relevant biochemical reaction databases.
• Network Visualization: Represents annotated mass peaks as a network where edges connect peaks whose candidate metabolites participate as substrates or products in known biochemical reactions.
• Cytoscape Export and Network Integration: Exports network visualizations to Cytoscape and permits overlay of graph-based analyses such as partial correlation networks.
• R Package Implementation: Provided as an implementation in R.

Scientific Applications:

• LC–MS Metabolite Annotation: Probabilistic assignment of candidate metabolites to observed mass peaks in LC–MS datasets.
• Metabolic Pathway and Network Interpretation: Linking annotated metabolites via biochemical reaction databases to interpret metabolic pathways and interactions.
• Systems Biology Network Analysis: Combining network outputs with Cytoscape and partial correlation networks for systems-level investigations of metabolomes.

Methodology:

Computational methods include Bayesian inference with Dirichlet-categorical priors, flexible likelihood-function calculations, construction of networks by linking peaks whose candidate metabolites act as substrates or products, and export of networks for Cytoscape-based analyses including partial correlation overlays.