Predição de microbioma saudável baseada em micro-organismos presentes no coral Mussismilia hispida, utilizando uma rede neural profunda

Abstract

One of the most diversified and productive ecosystems in the marine world are corals, providing, in addition to tourism, an important economic contribution to countries that have them on the coast. Thanks to genome sequencing techniques such as 16S sRNA, it is possible to identify the microorganisms that make up the coral microbiome, which play an important role in the health of the latter. The generation of large amounts of data thanks to the low cost of genome sequencing since 2005 offers a possibility for the use of artificial neural networks for the advancement of sciences such as biology and medicine. In this work, the prediction of healthy microbiome based on microorganisms present in the coral Mussismilia hispida collected in five reefs located near a marine protected area (“Parque Natural Municipal do Recife de Fora”) was performed, using a convolutional neural network and some classical machine learning algorithms such as SVM and decision tree, comparing their results obtained in several experiments.