This is a past event.
Michigan Technological University
Department of Biological Sciences
Biochemistry and Molecular Biology
Predicting Oil Contamination in the Great Lakes Using Microbial Communities
Abstract: Microbial community composition can shift dramatically in response to oil contamination. These microbial biomarkers can be identified by machine learning models in order to predict contamination and aid in environmental monitoring. 16S rRNA data from three different Great Lakes oil amendment microcosm studies was used to train machine learning models and explore their ability to predict oil contamination in the Great Lakes. The resulting random forest models were highly accurate, even when trained on as few as ten taxa. However, when testing the trained models with a wider variety of data from different times, locations, or sequencing runs, accuracy fell as the models lacked generalizability. Despite this, the accuracy of the models demonstrate the potential of microbial communities as biosensors able to identify trace amounts of oil and aid in determining spill origins.
Short Bio: Isaac is a Biochemistry and Molecular Biology PhD student in the Techtmann Lab with research focuses on microbial ecology and bioinformatics, specifically applying machine learning techniques to the modeling of environmental microbes.
0 people added