Banca de QUALIFICAÇÃO: CARLOS ALVES MOREIRA

Prediction of Protein Aggregation Regions based on Machine Learning and Tertiary Structure

   The average life expectancy of the world population grows gradually, accompanied by neurodegenerative diseases such as: Parkinson's, Alzheimer's, Amyotrophic Lateral Sclerosis, Type II Diabetes, Huntington's disease, and Spinal Muscular Atrophy. Currently, billions of dollars are spent on the treatment of these diseases, which highlights the importance and the need to invest efforts in the study and understanding of the molecular mechanisms involved. Despite a significant number of research conducted, these pathologies and the molecular mechanisms involved are not fully understood until now. Degenerative diseases usually involve some type of protein aggregation process, which are macromolecules involved in the biological process and whose function depends on their three-dimensional (3D) structure or native conformation. Studying the molecular mechanisms that lead to aggregation can help in the development of drugs and therapies. There are several studies that investigate the possibility of predicting regions prone to aggregation in the protein. Some of these works make use of computational methods based on experimental data. The use of Machine Learning (AM) techniques has been used quite successfully in the area of bioinformatics in the search for patterns, classifications and data predictions. This work has as main objective, the development of a tool to predict regions prone to protein aggregation, based on tertiary structure (3D) via Relative Surface Area (RSA), physical-chemical characteristics and using Machine Learning techniques. To carry out this work we will carry out: The study of the phenomenon of protein aggregation, physical-chemical characteristics, existing methods of prediction of protein aggregation propensity, available protein databases and Machine Learning (AM) algorithms. We propose the development of the MAGRE-II application focusing to  predict regions prone to protein aggregation using 3D structure and Machine Learning techniques. The results will be compared with the predictors found in the literature such as Aggrescan3D.