Condition-specific Antibiotic Regimen Assessment using Mechanistic Learning
Released via GPL GNU License
© 2022 The Regents of the University of Michigan
Chandrasekaran Research Group - https://systemsbiologylab.org/
Contact: [email protected]
This GitHub repository contains all code files used in data analysis and model development for the CARAMeL publication (citation below).
Chung, C. H., & Chandrasekaran, S. (2022). A flux-based machine learning model to simulate the impact of pathogen metabolic heterogeneity on drug interactions. PNAS Nexus, 1(3), 1–14. https://doi.org/10.1093/pnasnexus/pgac132
Successful implementation of the code files within this repository require the following external installations:
- MATLAB: preferrably version 2019b or higher
- Gurobi: optimization solver; preferrably latest version
- COBRA Toolbox: preferrably clone directly from GitHub
- Install a local version of this repository using the following command in a terminal:
git clone https://github.com/sriram-lab/CARAMeL.git
- Ensure to add the CARAMeL repository into your MATLAB path (see instructions here).
- You should now be able to run all MATLAB livescript files in the main repository folder.
The repository structure is outlined below. Of note, directories are prefaced by └─── while code/data files are preceded by │ .
CARAMeL └───code: contains all relevant code files │ └───CARAMeL_suite: contains all code files created to implement the CARAMeL approach │ │ │ caramel.m: main script the constructs a CARAMeL model │ │ │ caramel_anova.m: conducts a one-way ANOVA to determine GEM reactions with differential flux activity │ │ │ caramel_assessPerf.m: assesses the predictive performance of a CARAMeL model │ │ │ caramel_auroc: calculates the area under the receiver operating curve (AUROC) for a CARAMeL model │ │ │ caramel_classify: classifies drug interactions as synergistic, additive, or antagonistic │ │ │ caramel_crossValidation.m: conducts a k-fold cross-validation for a CARAMeL model │ │ │ caramel_features2gem.m: extracts GEM reactions associated with top CARAMeL model features │ │ │ caramel_featurize.m: implements featurization of raw input data into ML compatible format │ │ │ caramel_leaveOut.m: conducts a leave-out analysis for a CARAMeL model │ │ │ caramel_plot.m: generates plots visualizing CARAMeL model performance │ │ │ caramel_processInteractions.m: processes drug interaction data into a standardized format │ │ │ caramel_rankSubsystems.m: determines metabolic pathways enriched by GEM reactions tied to top CARAMeL features │ │ │ caramel_screen.m: screens all possible drug combination outcomes given a list of drugs │ │ │ caramel_topFeatures.m: extracts the top CARAMeL model features │ └───GEM_functions: contains all code files relevant to simulating metabolism using GEMs │ │ │ change_media.m: changes the simulated media condition for a given GEM │ │ │ constrain_flux_regulation.m: simulates reaction fluxes based on omics data constraints │ │ │ derive_flux.m: derives flux simulations for a specified list of conditions │ │ │ process_flux.m: processes simulated flux data into a format compatible with CARAMeL model construction │ └───misc: folder containing miscellaneous code files │ └────File_Exchange: code available through MATLAB File Exchange. Direct download links are provided for each method │ | └────confusion_matrix: determines the confusion matrix for two or more classes │ | └────hline_vline: generates horizontal and vertical lines on MATLAB plots │ | └────multiple_boxplot: generates multiple boxplots in a single MATLAB figure │ | └────permn.m: determines all possible permutations with repetition │ | └────progressbar: displays a progress bar to run a piece of code │ | └────raacampbell-sigstar-5aabaeb: plots statistical significance stars onto MATLAB plots │ | └────randomforest-matlab: Random Forests implementation in MATLAB │ └────custom: custom code files created for specific purposes. Refer to function file for details on input/output/usage │ | │ extract_from_confMatStats.m: extracts information from a confusion matrix │ | │ extract_metadata.m: extracts metadata from an Excel file │ | │ gscatter2.m: generates a 2D group scatter plot │ | │ meansgraph.m: customized modification to the built-in MATLAB function of the same name │ | │ process_chemgen_v2.m: processes chemogenomic data for E. coli (source) │ | │ process_transcriptome_tb.m: processes transcriptomic data for M. tb (source) │ | │ sig_boxplot.m: generates a multple boxplot figure with significance annotations │ └────optGpSampler_1.1_Matlab: code for optGpSampler │ └────populationFBA: code for population FBA └───data: contains all data files used for CARAMeL model development and downstream analyses │ └───supplementary: contains all supplementary data files │ │ │ ecoli_proteomics.xlsx: proteomics data for E. coli │ │ │ ecoli_transcriptomics.xlsx: transcriptomics data for E. coli │ │ GEM_info.xlsx: data pertaining to flux states simulated using the E. coli and M. tb GEMs │ │ biolog_PM01_conditions.xlsx: list of all conditions in the Biolog phenotype microarray 1 (PM01) (source) │ │ ecoli_chemogenomics.xlsx: chemogenomic data for E. coli (source) │ │ ecoli_interactions.xlsx: drug interaction data for E. coli │ │ ecoli_media.xlsx: details on different media conditions to simulate for the E. coli GEM │ │ iEK1008.mat: M. tb GEM (source) │ │ iJO1366.xml: E. coli GEM (source) │ │ mtb_interactions.xlsx: drug interaction data for M. tb │ │ mtb_transcriptomics.xlsx: transcriptomic data for M. tb (source) └───results: contains all results from CARAMeL model analyses │ │ CARAMeL_summary.xlsx: summary data from the E. coli CARAMeL model │ │ plot_data.xlsx: Excel file containing data pertinent for figure generation │ │ CARAMeL_workspace.mat: [REMOVED DUE TO SIZE LIMIT; ACCESS FILEHERE] MATLAB workspace file │ CARAMeL_figures.mlx: MATLAB livescript file that generates plots used in CARAMeL figures │ CARAMeL_flux.mlx: MATLAB livescript file that generates GEM flus simulation data for E. coli and M. tb CARAMeL models │ CARAMeL_logo.png: PNG file of the CARAMeL logo │ CARAMeL_main.mlx: MATLAB livescript file that conducts all main analyses for CARAMeL approach │ CARAMeL_validations.mlx: MATLAB livescript file that executes validations discussed in CARAMeL supplementary materials │ CARAMeL_visualization.Rmd: R markdown file generating additional plots used for CARAMeL figures │ LICENSE: LICENSE file │ README.md: README file │ populationFBA.mlx: MATLAB livescript file that executes code relating to cell-specific predictions