Author: Eric Hidari
Date: 30 December 2019
MATLAB model for ubiquitin chain reaction detailed in the paper: A general in vitro assay to study enzymatic activities of the ubiquitin system, Biochemistry, 2020 59 (7), 851-861, doi: 10.1021/acs.biochem.9b00602
USOS uses machine learning method to infer the rate constants in the ODE functions given the FRET measurement data. Namely, simulated annealing, an advanced Markov Chain Monte Carlo (MCMC) algorithm is applied here.
- MATLAB R2018b or higher
- MATLAB Global Optimization Toolbox, Parallel Computing Toolbox and Curve fitting Toolbox
Run Script_plot_model.m with the defined rate constants.
Run Script_simulation.m with the defined rate constants. Modify the code in lib/plot_model_conc.m to simulate concentrations of different species.
Run Script_optimise_rate_constants.m to infer rate constants from the data defined in raw_data folder. Change the epoch number in the script to run from multiple initial states.
The raw_data csv file is arranged as follows:
- The first column is the enzyme concentrations,
- The second column is the measured initial reaction rates,
- The third column is the standard deviation of the measured rates (unused).
The optimization algorithm simulated annealing can accept lower/upper bounds of the rate constants and the initial temperature as the parameters. These can be modified in lib/Vary_all_E_model.m
Alternatively, one can run Metropolis MCMC algorithm to observe the posterior distribution of the rate constants given the data in a Bayesian approach. Run Script_mcmc.m and change the chain length (nsimu) to generate a large number of posterior observations.
The Metropolis MCMC (mcmcstat) package is written by Marko Laine and is downloaded from: https://mjlaine.github.io/mcmcstat/
This software package is under the MIT License. See LICENSE file for details.