Evaluating Machine Learning Models for Essential Protein Identification

• Jessica da Silva Costa Centro Federal de Educação Tecnológica Celso Suckow da Fonseca (CEFET/RJ) https://orcid.org/0000-0002-0096-4147
• Jorge Gabriel Rodrigues Centro Federal de Educação Tecnológica Celso Suckow da Fonseca (CEFET/RJ) https://orcid.org/0000-0002-2027-7897
• Kele Belloze Centro Federal de Educação Tecnológica Celso Suckow da Fonseca (CEFET/RJ) https://orcid.org/0000-0001-6257-2520

Resumo

Drug development is often a complex and time-consuming process. Especially in the initial phase, selecting a target for drug development can take many years. Essential genes and proteins are biological entities responsible for the biological processes of survival and reproduction of organisms. Studies indicate that essential genes tend to have higher expression and encode proteins that engage in more protein-protein interactions. All these characteristics make essential proteins potential drug targets. Thus, this work proposes using protein-protein interaction-based features to train and evaluate machine learning algorithms to identify essential proteins. Experiments with the organism Saccharomyces cerevisiae indicate that the application of the Random Forest algorithm and balancing techniques obtained better recall values.