The discovery of life beyond Earth would be a watershed moment for astrobiology, but depending on what's out there, finding definitive evidence of extraterrestrial biology may not be as straightforward as "we'll know it when we see it." How can we reliably distinguish life from non-life, both on Earth and across the Solar System? And how can we hope to look for alien "lyfe" that differs markedly from life as we know it on Earth?
My group has pioneered a novel, robust method for distinguishing biotic from abiotic samples using a machine learning algorithm trained on pyrolysis–gas chromatography–mass spectrometry (py–GC–MS) data. We place an emphasis on explainable AI models, which means we can identify which molecular features are diagnostic of life and non-life. Our hope is that this method will help determine whether samples from other worlds contain traces of present or past life—even lyfe with non-Earthly biochemistry.
In the meantime, our AI-driven technique can be applied to ancient rocks on Earth to find traces of fossilized terrestrial biology to better our understanding of the evolution of life on our own planet. In our latest study, we find evidence for life in 3.33-billion-year-old rocks, including traces of photosynthetic life in sediments as old as 2.52 billion years.
Pyrolysis–gas chromatography–mass spectrometry data of an average bacterium from Cleaves et al. (2023). Figure by Michael L. Wong.
Further reading:
