Title of thesis project: Enrichment method for improving prediction of bioactivity in small molecules.
The increase in computational power and the amount of data that a researcher is supposed to process have contributed to the ever-increasing approach of computational methods for drug design. The first objective of scientists involved in computer-aided drug design is to effectively depict the structures of normal and pathological molecules which are then compared to pathogenic enzymes and active receptors respectively, as the purpose of pharmaceutical design is set. Thus, if the structure of a protein is known and the way in which the receptor or its active region acts, the model of action can be simulated, saving the time and cost that would be required by corresponding experimental testing.
In this diploma thesis, an extension of in silico predictions for bioactivity will be outlined, while a highly interpretable and repeatable model for pre-processing of in silico molecules is introduced, predicting the proteins correlated with psychosis disease. This workflow is transferable to other diseases.
Three python scripts were made for mining as well as manipulation of the data derived from DrugBank online drug database.