Installation instructions for Ubuntu 20.04.1LTS ( Linux 5.4.0-53-generic )

(Installation has only been tested on this system; Ryzen 3600 with 32GB )

1. Install Julia, ideally using the free version of JuliaPro (v1.4.0-1) https://juliacomputing.com/products/juliapro/ Alternatively download julia wget https://julialang-s3.julialang.org/bin/linux/x64/1.4/julia-1.4.2-linux-x86_64.tar.gz && tar zxvf julia-1.4.2-linux-x86_64.tar.gz

2. Install julia packages - either using the Pkg REPL - see https://docs.julialang.org/en/v1/stdlib/Pkg/ or through running julia add_packages.jl Required Julia packages are : ArgParse DifferentialEquations DiffEqCallbacks Random SparseArrays StatsBase Distributions Precise details of complete julia environment used listed in julia_installed_packages.txt

3. Install conda (via https://docs.conda.io/en/latest/miniconda.html ; or https://repo.anaconda.com/miniconda/Miniconda3-latest-Linux-x86_64.sh )

4. Create a new conda environment conda create -n crossover python=3.7 . This needs to be python3.7 to reproduce image analysis results correctly.

5. Activate this conda environment conda activate crossover

6. Install packages needed conda install matplotlib scipy statsmodels pandas lxml imageio click

7. Install pip-installable packages pip install svgutils dtoolbioimage ray

8. Install inkscape (needed for figure generation) sudo apt-get install inkscape

Alternatively to 4-7, the conda environment can be recreated using conda env create -f environment.yml (this specifies the precise versions of packages used by the authors)

To recreate results of manuscript

1. Run all simulations (takes several hours) python run_simulations.py . Modify my_env['JULIA_NUM_THREADS']='6' to the appropriate value for your machine (optional).

2. Dowload imaging data, placing these large files in a suitable location. For review purposes, these are hosted on Dropbox: https://www.dropbox.com/s/bugx89pwj8y598u/morgan_ox.tar.bz2?dl=0 (5GB) https://www.dropbox.com/s/y58jhqhxcq5as3v/morgan_ux.tar.bz2?dl=0 (3.5GB) https://www.dropbox.com/s/v2o7i142cy7abrg/morgan_wt.tar.bz2?dl=0 (16GB)

3. Decompress (bunzip2) and extract (tar xvf) these imaging files

4. Edit the data_path = 'xxxx' line in data_preprocess.py to point to this downloaded data

5. python data_preprocess.py (takes an hour or so).

6. Run bash all_plots.sh . Figures will be placed in output/figures folder.

To run short test of simulation code (c 1 min)

1. Ensure that Julia and the needed packages are installed (1. and 2. of Installation instructions)
2. julia new_sim.jl --n 2 (The julia executable may need to be added to your PATH).
3. Output is written to test0.1,2.1,360.0,30.0,0.0,false,0.5,1.25,1,100,6.8,1.2,1.1,2000,0.5,1.0,2.2,2.0,1.dat in the current directory Expected contents:

30.0
0.2372785017168222,6.284171195942124,6.329046064756079,7.549865490959156,8.435706968571894,9.381209050082024,10.395510425758815,12.741535514853943,13.113239238288754,14.658384902385036,16.67253261973717,23.19669145372131,28.55749019507202,29.59999100696399,29.
99713976695841
0.37283953307113954,0.37283953274230386,0.37283953273767434,0.37283953256574376,0.3728395324098933,0.37283953221189337,0.3728395319667844,142.744517700139,0.3728395320097407,0.3728395347390615,0.3728395381395296,0.3728395480086411,0.372839553939854,0.372839554
6504659,0.3728395547785809

This performs a single simulation, with the default parameters and the starting seed 1. Output consists of: l1 - the parameters used l2 - the length of the SC simulated l3 - the positions of the initial RI l4 - HEI10 levels at the RI at the end of the simulation

4. Parameter values and simulation results can be modified through command-line options: julia new_sim.jl --help for a complete list