Analysis code for "An online atlas of human plasma metabolite signatures of gut microbiome composition" by Dekkers et al.

Scripts

• correlations.R is the R-script used to calculate partial Spearman correlations between gut microbial species and plasma metabolites
• gsea.R is the R-script used to generate the enrichment test results for gut metabolic modules, taxonomic genera and metabolite classes
• variance_explained.R is the R-script used to calculate how much of the variance of gut microbial species is explained by a combination of metabolites
• create_test_data.R is the R-script used to generate the test data
• data.rda is simulated test data, the scripts work both on the real data as well as the test data
• Source_Data.zip is the data used to generate the manuscript figures

Installation guide

• Install R: https://www.r-project.org/

• Install Bioconductor: https://bioconductor.org/install/

• Install packages in R:

  • library(BiocManager)
  • install(c("rio", "BiocParallel", "ppcor", "glmnet", "fgsea"))

• Download correlations.R, gsea.R, variance_explained.R

• Install time: <10 minutes

Demo

• Download data.rda
• In R:
  • Set working directory to directory with data.rda, e.g. setwd("C:/Users/User/Documents/Demo/")
  • Run script correlations.R
  • Run script gsea.R
  • Run script variance_explained.R
• Expected output:
  • Files: correlations.tsv, diversity.tsv, gsea_module.tsv, gsea_genus.tsv, gsea_subclass.tsv and variance_explained.tsv
  • In correlations.tsv the strongest associations should be for metabolite1
  • In gsea_module.tsv and gsea_genus.tsv the strongest enrichment should be for metabolite1
  • In variance_explained.tsv the highest variance explained should be for metabolite1
• Run time: <5 minutes

Software versions tested: R 4.1.2, rio 0.5.29, BiocParallel 1.28.3, ppcor 1.1, glmnet 4.1-3 and fgsea 1.20.0