Tomasz Piotr Jurkowski

• Nowadays there are plenty of Computational Biology programs which are available to study various properties of proteins, like their structure, function, interactions with other proteins or evolution. In this work we have applied these tools to study RNA and DNA methyltransferases. Investigation of the DNA recognition by M.EcoRV aided with a homology model and sequence alignment shed light on the evolutionary pathways leading to change in sequence specificity of this protein, as well as on the mechanism and temporal order of events leading to the DNA recognition and methylation. Dnmt2 is an enzyme which looks like an authentic DNA methyltransferase but methylates tRNA molecules. This enzyme was studied using multiple sequence alignment and phylogenetic analysis as starting point. Our data clearly show that it follows the catalytic mechanism of m5C DNA MTases to methylate RNA. This represents the first example of such an enzyme. A homology model of the Dnmt3a catalytic domain, allowed us to study the function of its conserved residues and to interpret the experimental results. In addition, we have created a novel method to investigate protein-protein interaction sites, called "Absence of Interference" approach. We have employed this method to study the Dnmt3a/Dnmt3L complex. We have also contributed to the repository of bioinformatic programs, by creating new tools which are helpful to interpret and present experimental data. The Prima software is able to calculate conservation scores generated by the "Absence of Interference" method and map them onto protein models. DistriDist can analyze the distribution of pairwise distances between the query occurrences in a provided longer sequence, which can be helpful to identify the regions of the genome that can play a role in epigenetic regulation. BDPC was designed to analyze, present and compare the results from the bisulfite genomic sequencing projects, a laborious task which before had to be done manually.