Genetic Algorithm (2D-HP Protein Folding)

Genetic algorithm developed for the 2D-HP protein folding model. The application creates an initial population of conformations (a protein folding) and then applies selection, mutation and crossover.

Installation

The repository contains a Visual Studio 2013 project which includes all dependencies and can be run straight away.

General concept

• A conformation contains elements
• An element is either hydrophile or not
• An element has a position relative to the conformation
• An element has a direction pointing towards the successor's position (Left, Right, Straight)

Selection

The following selection methods are supported:

• Fitness-proportional
• Probability-based tournament
• Single Elimination tournament
• "Best wins" tournament

Tournament sizes (number of conformations) can be specified.

Logging

The algorithm automatically logs a runthrough to the average.txt file which contains the average energy of one population for each generation. This can be used to generate graphs like this (average energy vertical, generation number horizontal):

Additionally, the functions testTournaments and testRates are available, both of which use a set of parameters, do one runthrough of the algorithm. Then they change the parameters by a little bit and do another runthrough. This can be done for as many runthroughs as wanted. This way it's easy to analyze how the parameters affect the results.

Example

The easiest way to adjust the algorithm is by playing around with the parameters from the Algorithm::setUp method: setUp(int maxGeneration, int populationSize, std::string &chain, float mutationRate, float crossoverRate, Selection *selection, std::string& file = (std::string)"average.txt");

• maxGeneration: How many generations to iterate through
• populationSize: How many individuals a generation has
• chain: What benchmark sequence to use (SEQ20, SEQ36…)
• mutationRate: How many individuals are being mutated every generation (in %)
• crossoverRate: How many individuals are being crossovered every generation (in %)
• selection: What selection method to use
• file: Logfile for saving the values of each generation

An example call would be setUp(250, 5000, SEQ20, 0.08, 0.30f, new TournamentSelection(2.0, 80.0f));