RINALDO- RatIoNAL Drug design wOrkbench

Abstract

In the Pharmaceutical industry, namely in drug discovery, which is the first stage of rational drug design, much work is necessary to discover new medicines. Big pharmaceutical companies use data mining and machine learning tools to make this work faster and safer by using in-silico experiments. A way to improve the machine learning algorithms that aid in this discovery is by improving the datasets the algorithm uses as input; this is called Dataset Enrichment. An example of Dataset Enrichment is appending new data, adding attributes to the existing information present, thus increasing the variety of categories to which entry is related. In the context of drug discovery, information such as Taxonomies, Ontologies, and Knowledge Graphs play a significant role in providing a diversity of knowledge, consequently being of vital relevance to this enrichment. That is the focus of this dissertation. The proposed work consists of 2 parts: the completion of an initial prototype that already can perform machine learning experiments regarding adverse drug reactions between 2 drugs by adding a classification system to standardize the terms stored in the dataset; and secondly, the primary goal of creating a freely usable web application that allows for researchers to make drug-related experiments using public or user-inputted datasets, built on top of the initial prototype, taking advantage of the dataset enrichment applied. Finally, a case study will be conducted to compare the effectiveness and efficiency of the enrichment added.