Generative Artificial Intelligence accelerates solutions for reducing methane emissions and enhancing a sustainable future for animal agriculture

January 29, 2025 – Ames, Iowa – Methane emissions are a climate challenge to scientists and has led to 33% of greenhouse gas emissions and 3% of all greenhouse gases in the United States.

Ratul Chowdhury, Assistant Professor in the Department of Chemical and Biological Engineering and Black and Veatch Building a World of Difference Faculty Fellow at Iowa State University was recently featured in USDA ARS Research News USA and Dairy Reporter UK regarding work his lab is doing to reduce methane production in the livestock industry along with members of USDA ARS.

Previous studies have been done to identify fifteen molecules, like bromoform, that can act as natural inhibitors of methane production. These molecules are naturally occurring in the environment, like seaweed containing bromoform can be used in feed to reduce methane production. However, bromoform which achieves up to 90% methane reduction is a carcinogen and hence use of it is largely prohibited due to food safety and animal/ human health risks. The other molecules, while safe, have only showed ~30% methane reduction. This prompts research to find those molecules with similar inhibition efficiencies as bromoform that are non-toxic. The Chowdhury lab is leveraging AI to identify lead candidates from bovine-linked molecules (i.e., molecules that are already present in the cow’s body) which are very likely to be safe.

Chowdhury, a member of the Nanovaccine Institute, uses computational modeling to profile bovine metabolites and identify, using detailed molecular simulations (and experiments), the most promising candidates that can stop methane production in the cow’s stomach. These molecules are tested in laboratory for efficacy, food safety, and ease of production to remain agreeable to the agricultural market. Through multiple rounds of prediction reinforced with experiments, the team expects to disclose how and why the top performing molecules are able to reduce methane production in a cow’s rumen.

In a recent feature paper in Animal Frontiers Chowdhury and his USDA partners outline the molecular biology, AI, and experimental considerations both from a scientific and resource allocation point-of-view for such an undertaking. Chowdhury and the team pinpoint that generative AI integrated with structural molecular biology would be key to (a) understand how experimentally reported anti-methanogenic molecules interact with the key methyl-CoA-reductase enzyme in inhibiting methane production, (b) assess the metabolic role of microbial community in the cattle rumen, and (c) use them as priors to discover novel molecules. This technology not only could help solve greenhouse gas emissions from the livestock industry but also help tailor feeding programs for livestock to improve their diets, overall health, and income for producers.

The study is co-authored by Anthony N Frazier (USDA-ARS), Jacek A Koziel (USDA-ARS), Logan Thompson (Kansas State), Matthew R Beck (Texas A&M). For the first time, this study also includes a comprehensive economic analysis for such projects that involve experiments with animal feed diets. Analyses include computation and monetary breakdowns on a per molecule per cow basis to determine the expense and consequences of the research which, in turn, will steer investments to scale this laboratory research. Chowdhury’s team includes Iowa State graduate students Randy Aryee, Mohammed Sakib Noor, and postdoctoral fellow Arunraj B leading various aspects of this work.

Image courtesy: Ratul Chowdhury, see linked paper above.

Written by Pradeepa Sukumaran, Program Assistant II