Detection of mastitis-causing pathogen by sequencing different regions of 16S rRNA gene and machine learning

• Autor(a) principal: Clemente, Luan Gaspar
• Data de Publicação: 2022
• Tipo de documento: Dissertação
• Idioma: eng
• Título da fonte: Biblioteca Digital de Teses e Dissertações da USP
• Texto Completo: https://www.teses.usp.br/teses/disponiveis/11/11139/tde-16092022-110237/
• topic: 16S rRNA; 16S rRNA; Escherichia coli; Escherichia coli; Staphylococcus aureus; Staphylococcus aureus; Aprendizado de máquina; Bovine mastitis; Machine learning; Mastite bovina

The correct identification of mastitis-causing pathogens is a key factor in the successful management of dairy farms. Techniques such as culture medium, qPCR, and 16S rRNA sequencing have been used to detect important microorganisms in raw bovine milk samples. However, due to costs, some challenges remain. Machine learning methods have been shown as an attractive alternative, as they can integrate different sources of data, with a diversity of purposes. New studies focusing on the detection of clinical and subclinical mastitis highlight the potential of applied machine learning methods to the management of mastitis in dairy farms. In this work, we evaluate the performance of three machine learning methods to detect the most abundant mastitis-causing pathogen in individual raw milk bovine samples integrating data from milk composition and 16S rRNA sequencing. We show the potential for the identification of Escherichia coli and Staphylococcus aureus. For abundance greater than 3% in individual samples, an accuracy of 100% and 86% was achieved, respectively. These results show that not only subclinical and clinical mastitis can be detected by machine learning methods, but some mastitis- causing pathogens either. Moreover, to maximize the information obtained from 16S rRNA sequencing, we evaluate in silico genetic diversity for different regions of the 16S rRNAgene and validate the results by Illumina sequencing. We show that for better detection of microorganisms associated with bovine mastitis, the V2-V3 region detects a higher prevalence with more relative abundance. We hope that this work can contribute to better management of dairy farms as well as the development of new tools for the control of bovine mastitis.