This directory contains an automated pipeline for constructing cell-specific enzyme-constrained GEMs (ecGEMs) derived from Human-GEM (v1.3.0) based on transcriptomics and proteomics datasets.
- A functional Matlab installation (MATLAB 7.3 and higher)
- The RAVEN toolbox for MATLAB (version 2.3.0)
- For generating some figures, a functional R installation (version 3.6.1)
- The libSBML MATLAB API (version 5.13.0 is recommended)
- Gurobi Optimizer for any simulation
The ecGEMs are already present in the models/ subdirectory, but the scripts and data necessary to regenerate the models are available here. The master script for generating the ecGEMs from the tINIT GEMs (provided in the /models/humanGEM_cellLines/11models.mat file) is generate_human_ecModels_NCI60.m, located in the /ComplementaryScripts directory. Run this generate_human_ecModels_NCI60 script in MATLAB to regenerate the 11 ecGEMs.
Flux variability analysis (FVA) (corresponding to the results presented in Fig. 5B) can be run using the comparativeFVA_humanModels.m function in the ComplementaryScripts/Simulation subdirectory. Specify the name of the model (cell line) for which FVA is to be run; for example:
results = comparativeFVA_humanModels('HOP62');
To use the ecGEMs and non-ecGEMs to predict growth rates and metabolite exchange rates with increasing levels of constraints (as shown in Figs. 5C and 5D), run the predict_cellLines_gRates.m script in the ComplementaryScripts subdirectory.
The R script used to generate plots shown in Fig. 5 of the main text is plot_ecGEM_results.R, located in the ComplementaryScripts subdirectory.