Publicação

PhageDPO: phage depolymerase finder

Ver documento

Detalhes bibliográficos
Resumo:Antibiotic resistance is a severe public health problem. New resistance mechanisms are rapidly emerging and spreading globally, threatening our ability to treat infections. The bacteriophages (phages) arise as a possible solution through their capability of infecting and killing bacteria. Phages are natural bacterial predators: they encode an arsenal of specialized proteins to target their bacterial hosts. One emerging protein is Phages Depolymerases (DPOs), responsible for selective recognition and degradation of bacterial cell surface decorating polysaccharides, turning the bacteria susceptible to external agents. Due to the difficulty in locating these enzymes in the phage genome, we developed PhageDPO, a DPO prediction tool, through machine learning methods. Several classifiers were created, using different datasets and algorithms and tested through cross-validation. The datasets were composed of protein sequences retrieved from the NCBI protein database and by a different number of negative cases. Two models were selected for integration in the tool: the Support Vector Machine (SVM) model created with a dataset containing data of 4311 sequences and the Artificial Neural Network (ANN) model created with a dataset containing data of 7185 sequences. On an independent validation dataset, the SVM model presented 95% accuracy, 98% precision and 91% recall and the ANN model presented 98% accuracy, 99% precision and 96% recall. While the high precision and PECC of the SVM focus on predicting true DPO sequences and avoiding false positives, the ANN ensures that all DPOs are identified due to its high recall. PhageDPO was successfully tested in predicting DPOs of, previously characterized, phages. PhageDPO was integrated into the Galaxy framework (https://bit.ly/3dOam2u), providing a user-friendly graphical interface for wet-lab researchers without computational skills.
Autores principais:Duarte, José Alexandre Graça
Assunto:Bacteriophages Depolymerase Galaxy Machine learning Aprendizagem máquina Bacteriófagos Depolimerase Galaxy
Ano:2021
País:Portugal
Tipo de documento:dissertação de mestrado
Tipo de acesso:acesso aberto
Instituição associada:Universidade do Minho
Idioma:inglês
Origem:RepositóriUM - Universidade do Minho

Registos relacionados

Registos relacionados