Karakter seçimi metodunun moleküler ve morfolojik data kullanılarak filogenetik ağaç oluşturulmasında uygulanması

This thesis analyzes biological datasets in the form of (morphological data and molecular sequences data) by utilization of Genetic Algorithms as the Feature Selection method. Information contents of the datasets were investigated and discussed. The thesis also traced the congruencies and incongruences within the dataset and seek to answer questions such as what is the minimum number of characters required to reconstruct a desired tree, or which part of the dataset carries what type of information especially for taxonomic applications.This study is composed of two parts; analysis of morphological data and molecular sequence data. In the analysis of morphological data, I have analyzed a classical dataset based on Hoplitis taxa which can be accepted as a benchmark numerical taxonomy dataset and acorn dataset. The proposed method successfully optimized the Hoplitis dataset to a set of 67 characters which was very close to predictions of the authors of the dataset. Optimization of the acorn dataset was also quiet successful by returning reasonable characters.Shorter molecular sequences up to 1000 nt were also analyzed by the optimization algorithm. Heme Cupper Oxidase dataset were resulted with positions that are found within the solution subsets frequently. Some of those positions were highly accurate with the current literature. For the longer sequences, a similar algorithm has been designed by the use of Longest Common Substring method. In this method, only the difference is that analyses were performed on small subsets separated by common blocks rather than characters or residues. In this part, we have used a well-known mammalian mitochondrial genome dataset. The algorithm has returned a relatively small subset of data to construct the same phylogeny.