In the folder ComparingAlgorithms we present an analysis pipeline for comparing the three flux sampling algorithms (CHRR, OPTGP, and ACHR) available in the COBRA Toolbox (MATLAB) and COBRApy (Python).
We assess the algorithms' efficiency across three published biological networks: Arnold2014, Poolman2009, Gomes2010 as included in the folder.
We compare the three algorithms based on runtime (Sampling_OPTGP_ACHR.ipynb, Sampling_CHRR.m) and convergence (assessConvergence.m, ConvergenceDiagnostics.ipynb).
The assessConvergence.m assesses convergence in the sampling chains based on the interval-based potential scale reduction factor whereas the ConvergenceDiagnostics.ipynb assesses convergence in the sampling chains based on the Raftery and Lewis Diagnostic and the Gweke diagnostic.
We then constrained the Arnold2014 model with photosynthesis and metabolite data of plants grown at control conditions and of plants which had recevied one week of cold treatment at maturity. ExpData.xlsx contains the raw data which is analysed and summarized using the ExpData.R file. In-cabinet photosynthesis measurements and metabolites assays were done as outlined here.
Photosynthesis constraints were converted from area to FW measurements and scaled to a constant flux over an 8 h photoperiod. Constraints were applied to the genome-scale metabolic model of Arnold2014 as done in ColdAcclimationModels.ipynb and sampling was done using the CHRR algorithm (ArabidopsisSampling.m). Flux distributions were then plotted to a PDF file (FluxSamplingAnalysisArabidopsis.ipynb).
All scripts for this analysis can be found in the ArabidopsisStudy folder.
Our most recent Zenodo DOI generated for the journal submission is:
https://doi.org/10.5281/zenodo.3239075
If you are using our work, please cite us at:
Herrmann, H.A., Dyson, B.C., Vass, L. et al. Flux sampling is a powerful tool to study metabolism under changing environmental conditions. npj Syst Biol Appl 5, 32 (2019). https://doi.org/10.1038/s41540-019-0109-0