About

This is the GitHub repository for the manuscript Fine-tuning of energy levels regulates SUC2 via a SNF1-dependent feedback loop [1]. This repository contains all the code required for reproducing the modelling results presented in the paper.

The aim of the modelling was to elucidate the mechanism behind recently collected single-cell time-lapse data of the SUC2-gene expression upon long term glucose starvation (Fig. 1) in Saccharomyces cerevisiae. To model the single-cell time-lapse data in an ODE-framework, two methods were used and compare. The standard two stage (STS) approach [2], and non-linear mixed effects (NLME) modelling [3]. What causes the decrease in SUC2-expression (Fig. 1) is currently unknown. In the manuscript we propose, by combining dynamic modelling and experimental data, a mechanism for this decrease.

Figure 1: Single cell SUC2 gene expression measured via YFP in S. cerevisiae when the external glucose is reduced from 4 % to 0.1 % at time zero. The black lines corresponds to the observed 0.05, 0.5 and 0.95 quantiles, while the blue line are four randomly chosen cells.

To fully replicate the result presented in the manuscript the requirements in the Requirements for replication of result section should be fulfilled. Given this, the result should be reproducible by running the Run_all.sh-script, which can be found in the Code directory.

Repository structure

Each directory contains a README.md describing the role of that directory. These roles can be summarised as:

• Data: Contains the original SUC2 data in CSV-format. Upon cloning the project the data is not in the folder, but it can be retreived from here. Note, the data must be placed in Data directory for the code (Run_all-script) to work.
• Intermediate: Contains intermediate which files are produced by the analysis. Files here are not counted as results, but are required for producing the figures and tables in the result directory.
• Code: Contains the code for processing the data, running STS and NLME, and plotting the end result. The Monolix files, e.g mlxtran-files, are also located here.
• Result: Contains, after running Run_all.sh, the figures presented in the manuscript. It also contains tables with estimated parameter values for the NLME and STS approaches.

Requirements for reproducing the result

This section contains information of operating system, programming languages (with libraries) and software required to reproduce the result.

Operating system

This repository was created on a computer running on Ubuntu 18.04.4 (Linux). In the code, all file-paths were encoded as relative paths using Unix-paths (e.g dir_data = ./../Data/). The run-all script is a shell-script (bash). Hence, the code should be able to run on any Unix-based system (e.g Linux, Mac). However, problems can arise with the file-paths and shell-script on Windows. One way to resolve this for a Windows user might be (I have not tested this) to download the Ubuntu-terminal for Windows. If this fails, a Windows user can change the paths in the scripts and runs the files in the order described in the README-file in the Code-directory.

Programming languaes (and libraries)

Three programming languages are required for reproducing the results, Julia (version 1.3.1), Python (version 3.7.4) and R (version 3.6.3)

Julia (version 1.3.1) [4] was used for STS approach, and for simulating cells given estimated population parameters. As Julia is language under heavy development, reproducing the results might fail if not the exact same Julia and library versions are used as those listed here. The libraries used are:

• CSV (version 0.5.24)
• Distributions (version 0.22.4)
• NLopt (version 0.5.1)
• ProgressMeter (version 1.2.0)
• DifferentialEquations (version 6.11.0)
• DataFrames (version 0.20.2)

R (version 3.6.3) was used for processing the data and plotting. Mostly the tidyverse was used for this [5], and as only standard functions were used, the result should be reproducible if newer (but also older to a limit) versions of R and the listed libraries are used. The libraries used are:

• tidyverse (version 1.3.0)
• stringr (version 1.4.0)
• latex2exp (version 0.4.0)
• ggthemes (version 4.2.0)
• readxl (version 1.3.1)

Python (version 3.7.4) was used to process parts of the Monolix result. As only standard functions were used, the result should be reproducible if newer (but also older to a limit) versions of Python and the listed libraries are used. The libraries used are:

• numpy (version 1.17.2)
• pandas (version 0.25.1)

Software

Monolix (version 2019R2) [6] was used for the NLME-parameter estimation. I do not know how much changes between versions, hence I recommend using the exact same version if attempting to reproduce the results. When running the parameter estimation, the default seed in Monolix was used.

References

1. Persson S, Welkenhysen N, Shashkova S, Cvijovic M. Fine-tuning of energy levels regulates SUC2 via a SNF1-dependent feedback loop. Front. Physiol. 2020; 11:954
2. Karlsson M, Janzén DL, Durrieu L, Colman-Lerner A, Kjellsson MC, Cedersund G. Nonlinear mixed-effects modelling for single cell estimation: when, why, and how to use it. BMC systemsbiology. 2015;9(1):52.
3. Davidian M, Giltinan DM. Nonlinear models for repeated measurement data: an overview and update. Journal of agricultural, biological, and environmental statistics. 2003;8(4):387.
4. Bezanson J, Edelman A, Karpinski S, Shah VB. Julia: A Fresh Approach to Numerical Computing. SIAM. 2017;59(1):65–98.
5. Grolemund G, Wickham H. R for data science. 2018.
6. Monolix version 2019R2. Antony, France: Lixoft SAS; 2019. http://lixoft.com/products/monolix/.