Chapter 11 - Evolutionary Computation and Fractal Visualization of Sequence Data

Abstract

This chapter explores a particular class of visualization techniques for sequence data. Each of the techniques discussed results in fractals, which are well known for their complexity and visual beauty. Evolutionary computation is used in this chapter to optimize the extent to which the fractals produced are able to separate different categories of input sequence data. The overall method has merit because fractals are useful for conveying many different types of information within one picture by using both shape and color. It first describes a standard technique for visualizing DNA or other sequence data with a fractal algorithm, and then generalizes this technique in two different ways to obtain two new types of evolvable fractals, including an indexed IFS, uses incoming sequence data to choose which contraction map will be applied next, and chaos automaton that drives a finite-state machine. Both of these new methods are forms of iterated function systems (IFS), which are collections of randomly driven or data-driven contraction maps. Both evolvable fractals are tested on their ability to visually distinguish among different DNA from distinct microbial genomes. Finally, the chapter discusses potential improvements in the fractal chromosomes, fitness functions, and various other issues that must be resolved to obtain useful applications.