Jeremy R. Ash, Melaine A. Kuenemann, Daniel Rotroff, Alison Motsinger-Reif, and Denis Fourches
To run any of the r scripts provided, clone the repository and open the metabochem.Rproj file in Rstudio. Run the scripts within the project, so that they can find the relevant paths on your machine.
- The analyses folder contains three rmarkdown files and one r script, which perform the majority of the analyses reported in the paper.
- differential_analyses.rmd: Performs the differential analysis on cancer and healthy patient metabolites. Outputs the significance results. See already rendered report: differential_analysis.pdf.
- metab_clus.r: Filters metabolites by significance and clusters on chemical structure. Outputs clustered metabolite concentration profiles.
- metab_classifier_plasma.rmd: Takes the clustered plasma metabolite concentration profiles as input. Builds and validates classification models predicting patient cancer status. See already rendered report: metab_classifier_plasma.html.
- metab_classifier_serum.rmd: Takes the clustered serum metabolite concentration profiles as input. Builds and validates classification models predicting patient cancer status. See already rendered report: metab_classifier_serum.html.
- The analyses/data folder contains the data used by each analysis script.
- sample_metabolites_training_excol_fix.csv, sample_metabolites_test.txt: raw patient metabolite concentration profiles.
- training_set_tq_normalize.rda, test_set_tq_normalize.rda: fully processed and normalized patient metabolite concentration profiles.
- sample_factors_train.txt, ssample_factors_test.txt: patient characteristic data, including cancer status.
- common_test_training_molecule-v2.sdf: metabolite chemical structures.
- metabolics_fingerprint.csv: metabolite fingerprints.
- ML_data.RDATA: clustered metabolites concentration profiles.
- The analyses/data folder/signficance_results folder contains the signficance_results from the differential analysis.
- The analyses/figure_scripts folder contains the scripts to recreate a few of the figures in the paper. The file names contain the corresponding figure numbers.