Computational protocol: Exercise and high-fat feeding remodel transcript-metabolite interactive networks in mouse skeletal muscle

Similar protocols

Protocol publication

[…] To perform an integrated analysis of the transcriptomic and metabolomic data, we generalised Weighted Gene Co-expression Network Analysis. First, missing values in the metabolomics data were imputed using k-nearest neighbours imputation with k = 3. Second, to reduce the dimensionality of the final matrix only the top 20% of data (by variance) for each dataset (metabolomics from positive ion spectroscopy, from negative ion spectroscopy and RNA-Seq) were carried forwards. These data matrices were combined, mean centred and z-scaled. Finally, WGCNA was carried out as described elsewhere. Modules of correlated analytes (metabolites and transcripts) were identified using a measure of topological overlap and each named with a colour for easy reference. For each module, a module eigengene was computed (defined as the first principal component). This eigengene provided an aggregate measure for all the analytes in the module and was used to identify modules of interest (i.e. those that were affected by experimental interventions). Finally, seven modules of interest identified thus were passed into MetaCore (Thompson Reuters) for enrichment analysis and pathway mapping. All analyses prior to MetaCore were conducted in R; code is available on request. […]

Pipeline specifications

Software tools WGCNA, MetaCore
Application RNA-seq analysis
Organisms Mus musculus