Neural Network Modeling of Species Evolution: Loss and Gain in Complete Genome

The recent achievements in sequencing of the genomes of living organisms result in the emergence of extensive amount of DNA sequences. The novelty of these data offers a unique opportunity to study the origin and evolution of life. Every genome comprises thousands of genes and their complement can be used to analyze the major aspects of organism biology. The majority of genes is inherited from the ancestor and can be assembled into phylogenetic clusters of orthologous groups (COG). Neural networks appear to be practicable for further analysis of the clusters. Proposed is the innovative neural network topology capitalizing tree-like structure with back propagation for the phylogenetic analysis of complete genomes. It is used to estimate the probabilities of gene loss and gain along the branches of the given phylogenetic tree. The validity of neural network calculations was verified by the results of the simulated evolution process. The confirmed model was used for the sample of 12 recently sequenced genomes from the taxonomic class of gamma-Proteobacteria and 12 genomes from super-kingdom Archaea.