Prediction of host phenotypic outcomes with Machine learning

Description:
To reduce the environmental impact and costs for meat production and maintain the feeding standards of an ever-growing human population are some of the major concerns of the beef cattle industry worldwide. While several improvement efforts have been performed to tackle this problem, including genomic selection, it was not until recently that the microbiome started to be considered as an important source of phenotypic variation in ruminants. The increasing evidence of the microbiome role in health, development, and environmental impact of beef cattle, leverages the microbiome as an interesting research object for data analysis and Machine Learning since the identification and manipulation of patterns of microorganisms in farm animals could help mitigate costs and increase their overall efficiency. The microbiome field relies on approaches that randomly sequence the genetic material (DNA or RNA), generating massive, sparse, and sometimes compositional data, being this field an example of big data exploration in biological sciences. The search for patterns that are associated with environmental features and can be used to predict phenotypic outcomes is challenging and could be benefited from data-driven investigations, such as Machine Learning and Deep learning approaches. This research aims to explore the faecal microbiome of 313 Angus Calves and to identify the relation between microbiome features and host’s phenotypic outcomes, like growth and health status. Altogether, the results generated by this project have the potential to significantly increase knowledge related to beef cattle, paving the way for a new layer of information to be considered in animal breeding and production. It also has the potential to develop and validate methods that might be useful for other biological models.