Natural Selection Model

A model for evolution by natural selection that makes many simplifications.

This small project was undertaken to assist my understanding of object-oriented programming in Python.

Developed: January 2021

Running

Set permissions, run program:

$ chmod 755 main.py
$ python3 main.py

Demonstration

Initial Population Subset

[TCTGGTCCCACACGTT] Gen:0
[GCTCGGCTACAGCCGG] Gen:0
[ACAAGGTGGGCAAGAA] Gen:0
[ATGGGTCCCCCGGGGC] Gen:0

Population Subset (after 30 generations)

[GGGGGGGGGGGGGGGG] Gen:30
[GGGGGGGGGGGGGGGG] Gen:30
[GGGGGGGGGGGGGGGG] Gen:30
[GGGGGGGGGGGGGGGG] Gen:30

(With optimal DNA set to 16 Gs)

---

Modelling: Genetics

• DNA simplified to 16 single nucleobases (A,C,G,T)
• Assumes a being's phenotype is completely controlled by its genetics
  • This makes evaluating fitness easier

Modelling: Heritable Variation, Differential Reproduction

• Each being in the initial population has randomly set genetics
• Heritability is modelled via reproduction:
  • Each base of the offspring DNA is picked randomly between both parents
• Differential reproduction:
  • Beings with reproductive advantages have a higher fitness rating
  • Higher fitness correlates to higher rates of survival
  • Which, on average, results in differential reproduction

Random Mutation

• At creation, each being can be subject to random mutations:
  • Each base in the genome has a small chance of randomly mutating
    (one in mutation_rate_denominator (set to 1,000) chance of occuring)
  • A random mutation occurs, on average, once in a set of 32 beings
  • Naturally, these mutations are heritable, and affect fitness assessment

Modelling: Fitness

• Fitness is represented as a probability: 0 for unfit, 1 for optimally fit
• Fitness is calculated using the get_survival_probability() method
  • closeness of the being's genetics to the specified optimal genetics considered (0-1)
  • result is scaled using the exponential function:
    survival_prob = closeness^(exponent), exp >= 1
  • the higher the exponent, the more selective survivability is (as fitness is graded lower)
  • optimal beings will always be graded optimally fit using this model
• Beings greater than 4 generations old are set to unfit, and don't progress to generation 5

Modelling: Competition

Controlling Population Size with a Dynamic Environment

• Due to computational limitations, exponential population growth must be avoided
  • Environment is dynamically changed to impose greater selection pressures on population
  • (Simulates resource scarcity from increased competition, increasing selection pressures)
  • critically high population count is stored in critical_high_pop, initially set to 10,000
• To prevent rapid population decline (and simulation termination):
  • Environment changed to impose less strict selection pressures
  • (Simulates resource abundance from reduced competition, decreasing selection pressures)
  • Critically low population count is stored in critically_low_pop, initially set to 500
• This functionality is poor, and must be improved with further research