The Genetic Algorithm for biologists

Abstract

Basic scripting and programming skills are becoming increasingly important for scientists. Optimization problems are common in biology, for example to predict how proteins or RNA will interact with each other or with small molecules. Sometimes one needs to optimize many variables at once, using complex fitness functions or criteria that may have many local minima. One established and powerful optimization technique is the genetic algorithm. The genetic algorithm organizes parameters, called “genes,” into “chromosomes,” and creates a population of such chromosomes which can “breed” with each other and acquire mutations. Breeding is a cool idea because in principle two individuals could have optimized different parts of their chromosomes, and their offspring could have the advantages of both “parents.” I will discuss the concepts of population, crossover, mutation, convergence, and more. We will discuss the example of docking, for which the genetic algorithm has been used successfully. Lastly, we will talk about results and future directions in computational protein engineering, in which we may use the genetic algorithm to greatly increase protein-protein binding affinity.