Characterization and annotation of archaeal promoter sequences

Abstract

Archaea were not long ago included as a particular domain in the tree of life. This vast and unexplored domain requires a plethora of techniques aimed to produce reliable inferences on it since not much information has been annotated. The objective of this thesis is to create a bioinformatics pipeline that encounters promoters in unannotated genomes of archaea. To perform such task, the conservation reported in the binding site of specific transcription factor proteins is considered in order to draw a profile of promoter-like sequences derived from experimentally validated promoters of the archaea H. volcanii, S. solfataricus, and T. kodakarensis. Additionally, the conserved aspect of these organisms is employed is a classification task based on statistics and artificial neural networks. The results obtained in this thesis are displayed in the form of two scientific articles. Where in the first, the binding site of transcription factor proteins that assist RNAP binding was located even in promoter sequences that lack the canonical binding sites as well as unannotated organisms. Secondly, a combination of statistics and artificial neural network classification has succeeded in classifying archaeal promoters in a rate > 90%, which has shown a satisfactory way to deliver annotation upon sequences whose genome is not fully explored. [resumo fornecido pelo autor]