BY MARIA BURKE
Combining state-of-the-art chemical techniques with artificial intelligence (AI), an international team of researchers has uncovered the faint chemical echoes left by living organisms inside ancient rocks that formed over 3.3bn years ago. The same methods could be applied to samples from Mars or other planetary bodies to determine whether they once supported life.
Life on early Earth left behind little molecular evidence. Fragile materials such as primitive cells and microbial mats were squashed, fractured and heated as the planet’s crust shifted over billions of years. These intense processes destroyed most original biosignatures that could have provided insight into life’s earliest stages. This means teasing out biochemical information from ancient organic-rich sediments, such as when photosynthesis emerged relative to the inferred oxygenation of Earth’s atmosphere, remains a challenge.
‘Ancient rocks are full of interesting puzzles that tell us the story of life on Earth, but a few of the pieces are always missing,’ says Katie Maloney of Michigan State University, US. ‘Pairing chemical analysis and machine learning has revealed biological clues about ancient life that were previously invisible.’
‘Ancient life leaves more than fossils; it leaves chemical echoes,’ agrees co-lead author Robert Hazen from the Carnegie Institution for Science in Washington, DC, US. ‘Using machine learning, we can now reliably interpret these echoes for the first time.’
The team used high-resolution pyrolysis gas chromatography and mass spectrometry (py–GC–MS) to break down both organic and inorganic material into molecular fragments. They then trained an AI system to recognise the chemical ‘fingerprints’ associated with biological origins.
The researchers analysed more than 400 samples, ranging from modern plants and animals to billion-year-old fossils and meteorites. The AI system distinguished biological from nonbiological materials with over 90% accuracy and detected signs of photosynthesis in rocks at least 2.5bn years old (M. Wong, et al, PNAS, 2025, DOI: 10.1073/pnas.2514534122).
Before this work, dependable molecular evidence for life had only been identified in rocks younger than 1.7bn years. This new approach effectively doubles the period during which scientists can study chemical biosignatures.
The researchers say they have yet to fully exploit the opportunities provided by machine learning. In future work, they plan to combine py–GC–MS analyses with other analytical techniques including Raman and Fourier Transform IR spectra, and other data such as the stable isotope ratios of carbon, hydrogen, oxygen, nitrogen and sulphur.
Visit Blog
