On 26 August 2019 at 2.15 p.m., in J.Liivi str. 2 room 405, Tõnis Tasa will defend his thesis "Bioinformatics approaches in personalized pharmacotherapy" for obtaining the degree of Doctor of Philosophy (Computer Science).
Prof. Jaak Vilo (Institute of Computer Science, UT)
Assoc. Prof. Tuuli Metsvaht (Institute of Clinical Medicine, UT)
Res. Prof. Lili Milani (Institute of Genomics, UT)
Prof. Inge Jonassen (University of Bergen, Norway),
Prof. William Hope (University of Liverpool, UK).
The amount of collected health data is growing fast. Insights from these data allow using biological patient specifics to improve therapy management with further individualization.
This thesis addresses problems in multiple sub-fields of personalised medicine and aims to illustrate that data for precision medicine emerges from different sources.
Drug metabolism is difficult to predict because individual biological differences. Fortunately, drug concentrations are a good proxy for drug effect. To address the growing need for tools that allow on-line therapy adjustment based on individual concentrations we have developed and externally evaluated a precision dosing tool that allows individualised dosing of vancomycin in neonates.
Other than drugs used in therapeutic drug monitoring, most pharmacotherapies can not rely on continuous concentration measurements but for such drugs genetics provides a valuable source of information for individualization. Effects of many genetic variants in drug metabolism pathways are often large enough to require changes in drug prescriptions or chedules. We have applied a population-based approach in testing relations between drug related adverse effects and genomic loci, and found and validated a novel variant in CTNNA3 gene that increases adverse drug effects in patients with oxicam prescriptions. This was done by leveraging the data in Estonian Genome Center and linking these to nation-wide electronic health data registries.
Computational genetics relies on quantitative methods for which the most common is the genome-wide association analysis (GWAS). A common GWAS downstream step involves time-consuming visual assessment of the association study p-values in context with other variants in genomic vicinity. In order to streamline this step, we developed, Manhattan Harvester and Cropper, that allow for automated detection of peak areas and assign scores by emulating human evaluators.