Spinocerebellar Ataxia type 3 (SCA3) is a rare and progressive neurodegenerative disease, caused by a mutation in Ataxin-3 protein, and particularly on its polyglutamine tract. Its in-trinsically disordered domain and complex structural dynamic make it difficult to find new molecules that work on the protein, which in turn makes it challenging for researchers to stop disease progression. In order to reduce costs and time spent on discovering new benefi-cial molecules, my thesis seeks to develop and optimize a 3D property-based Ligand-Based Virtual Screening (LBVS) pipeline. The pipeline I developed is based on the CHEESE framework, an algorithm that was recently developed to identify new molecules in large li-braries using ShapeSim and EspSim algorithms, with electrostatic and shape similarity as key features.