polyomino-evolution

Simulation of evolution and complexity in polymino assembly model -- as in Johnston et al. "Symmetry and simplicity spontaneously emerge from the algorithmic nature of evolution" Proc Natl Acad Sci USA 119 (11) e2113883119 (2022) here https://www.pnas.org/doi/10.1073/pnas.2113883119

Requirements: C, R with ggplot2 , Bash for the wrapper script

run-blend.sh is wrapper code, which compiles and runs a set of experiments in Bash

sim-poly-blend.c wraps an individual experiment. It takes command-line arguments (with defaults):

• --directed 1 (directed evolution? 0 no, 1 size-based fitness, -1 random fitness)
• --npar 10 (number of individuals in evolutionary population)
• --mut 0.1000 (mutation rate per genome bit)
• --targetsize 16 (target size for size-based fitness function)
• --ntile 16 (number of tiles in an assembly genome)
• --ncol 64 (number of colours in an assembly genome)
• --numr 5000 (number of evolutionary runs)
• --nsamp 100000000 (number of samples of genotype space)
• --outputall 0 (output all structures found, or just top-fitness ones? 0 just top fitness, 1 all)
• --confound 0 (apply a confounding step to genomes? 0 no, 1 cyclic permutation based on 1s count, >= 2 random changes with prob (argument-1)/(genome length))
• --rseed 1 (random seed)

The outputs, depending on the experiment, are files labelled as follows:

• out- -- general output, including structures discovered, bitstring and reduced genomes, and evolutionary timescales
• stats- -- (CSV) statistics of recorded structures, including size, symmetry, modularity, various complexity measures, frequencies in evolution and sampling
• lib- -- shapes of recorded structures, as ASCII characters on a square grid (. denotes an empty site)
• pop- -- population structure over time and simulation instances, with references referring to the structures in stats- and lib- (-1 denotes an unbound or non-deterministic individual)

The other .c files take care of things under the hood:

• assembly.c -- polyomino assembly
• ga.c -- genetic algorithm
• library.c -- structure comparison and storage
• stats-variable.c -- various summary and complexity statistics for polyomino structures

Visualisation is done in R:

• plot-blend.R plots various summaries of the outputs, corresponding roughly to manuscript figures
• plot-structs.R , passed a command-line reference to a particular experiment, draws the top polyomino structures from that experiment
• plot-complexity-hists.R plots complexity histograms over (simulation) time
• (files with prepended nz- produce these plots using the nnonzero (number of nonzero genome sites) complexity measure instead of the ninterface (number of interface types)