Proposta de uma GAN para geração de moléculas

Abstract

In the field of biology, molecules play a crucial role, acting as the fundamental building blocks that make up everything around us, from elements like water and air to the intricate structures that make up our own bodies. Each molecule is made up of atoms, which represent the fundamental elements of chemistry. Molecule synthesis is a fascinating process that involves creating new configurations of atoms to create specific molecules. This approach is extremely relevant, since different molecules exhibit different properties. Some of these molecules play an essential role in treating diseases, while others can be incorporated into resistant materials or transformed into energy sources. The growing interconnectivity between medical studies and information technology for the construction of efficient and precise tools for pharmacological innovation is evident. This work proposes a generative model that uses Ensemble Learning strategies applied in a cGAN (Conditional Generative Adversarial Network) to generate new candidate chemical compounds for medicines. The proposed approach has several potential applications in drug discovery, including identifying new therapeutics and optimizing lead compounds, and opens scope for future research.