Microbiome of the deep sea
Beyond her hands-on lab work, David is pioneering artificial intelligence applications in microbiome research. By training machine learning models on massive datasets, her team is discovering how to predict patterns and identify microbial signatures linked to different conditions.
Her AI approach functions similarly to how a person might read thousands of books to develop a deep understanding of a subject before applying that knowledge to something new. Instead of analyzing each microbiome sample from scratch, her team feeds AI models vast amounts of microbial sequencing data, allowing the system to learn and recognize relationships between the different microbes. These models can then be applied to help classify conditions such as inflammatory bowel disease or colorectal cancer with greater accuracy.
“It is awesome, because the model can remember relationships that us humans might not. It’s finding these complex patterns,” David said.
One of the major challenges in microbiome research is the sheer volume of data involved. Each individual has a unique microbiome comprising thousands of different microbial species, each interacting in complex ways. Traditional methods of analyzing these communities can be time-consuming and require extensive resources. AI provides a way to quickly process and interpret large datasets, identifying patterns that can reveal valuable insights.
Her latest National Science Foundation study continues to push the limits of what AI can do. With a $540K grant, David is applying deep learning to analyze oceanic microbial ecosystems, an extension of her expertise in microbiome research.
The deep sea is a crucial, yet poorly understood driver of global biogeochemical cycles, the movement of essential elements like methane and nitrogen. These cycles regulate ecosystem function, influence climate and support life.
“We are looking at microbes in the ocean and researching how we can use AI to discover what role unknown genes play in methane seeps off the coast of Oregon and Washington,” she said.
Methane seep habitats, areas where methane gas escapes from the sea floor, are unique, diverse areas nourished by methane-consuming microbes. However, many of the genes involved in these deep-sea cycles remain unidentified, limiting our understanding of how these ecosystems function and their impact on global biogeochemical processes.
To analyze these complex environments, researchers will develop two AI models designed to decode gene functions. The first model will categorize genes into pathways by studying how they appear together in microbial communities. The second will use generative AI to predict the functions of unknown genes based on protein sequences and text-based data. Together, these models will help scientists identify genes responsible for each of the cycles identified.
The main outcome will be a scalable approach to artificial intelligence that will advance key questions in earth system science. Understanding the genetic mechanisms behind biogeochemical processes is crucial for predicting how ocean ecosystems respond to environmental changes.
The results of this study will include exhibits by artists involved in the research as well as a documentary about how AI can harness big data to help advance the understanding of earth systems.
As science continues to reveal the hidden influence of the microbiome, one thing is clear: critical solutions lie in understanding the powerful role microorganisms play in our bodies and our environment. David’s research has us on the right path to new understandings.