AI Uncovers Oldest-Ever Molecular Evidence of Photosynthesis

AI discovers oldest molecular evidence for photosynthesis

By Lee Billings edited by Claire Cameron

Although much of the history of life on Earth is written, the first few chapters are obscure to say the least. In our ever-changing world, the older a rock is, the more it has changed, obscuring or even erasing traces of ancient life. In fact, beyond a fuzzy boundary of about two billion years, this interference is so complete that no intact, intact Earth rocks are known, making any potential signs of biology as clear as mud.

At least until now. In a study published on November 17 in the Proceedings of the National Academy of Sciences, A group of researchers say they have harnessed artificial intelligence to trace life further back in time than ever before, using machine learning to distinguish echoes of biology from simple abiotic organic molecules found in 3.3 billion-year-old rocks.

The results could more than double the date when scientists can convincingly claim to discern molecular signs of life in ancient rocks, the study authors say, citing previous record-breaking measurements involving rocks 1.6 billion years old.

On supporting science journalism

If you enjoy this article, please consider supporting our award-winning journalism by subscription. By purchasing a subscription, you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.

The study also reveals signs of photosynthesis in rocks 2.5 billion years old, some 800 million years earlier than any other confirmed biomolecular evidence. The authors suggest that in the not-so-distant future, similar techniques could be used to search for signs of extraterrestrial life on Mars or on the icy oceanic moons of the outer solar system.

And such astrobiological applications would not necessarily require the extremely expensive task of retrieving material from Mars or any other extraterrestrial location for further study in laboratories on Earth. “Our approach could be carried aboard a rover, without the need to send samples home,” explains the study’s first author, Michael Wong, an astrobiologist at the Carnegie Institution for Science.

According to Karen Lloyd, a biogeochemist at the University of Southern California who was not involved in the study, the technique holds promise as an “agnostic” way to search for life, independent of Earth-related hypotheses.

“This allows for possible extrapolation from an extremely varied and diverse data set of biomolecules present in known living matter, extending to matter that may or may not originate from living things,” says Lloyd. “This is really useful in the search for life on rocks from ancient Earth, as well as rocks from extraterrestrial bodies.”

Rocks containing familiar fossils – dinosaurs, ferns, fish, trilobites, etc. – may seem extremely ancient, but they actually represent less than 10% of Earth’s 4.5 billion years of history. In other words, for each of the 500 million years that make up the Phanerozoic Aeon (Greek for “visible life”), there is nearly a decade of underlying planetary time during which early life flourished almost imperceptibly, barely registering in the fossil record beyond traces of molecules such as lipids and amino acids.

The problem, says lead author and Carnegie geologist Robert Hazen, is that these molecules degrade and disappear over time. “Our method instead looks for patterns, such as facial recognition of molecular fragments,” he explains. “Think of the burned Herculaneum scrolls that AI helped “read.” You and I only see dots and squiggles, but AI can reconstruct letters and words.

The team began by collecting more than 400 samples, some modern, some ancient, some from known abiotic sources like meteorites, others filled with fossils or living microbes, and several containing organic molecules but no obvious indicators of life. They fed them into an instrument called a pyrolysis gas chromatograph mass spectrometer (Py-GC-MS), which vaporized each sample to release and then classify their constituent molecular fragments based on their mass and other properties. This resulted in a rich “chemical landscape” for each sample, filled with tens of thousands or even hundreds of thousands of peaks, denoting different possible compounds, and ready for the AI ​​to examine closely.

After training the AI ​​on about 75% of the data samples, the researchers released it on the remaining 25%. The system correctly distinguished between biotic and abiotic samples for more than 90% of this material, but its certainty decreased as a rock’s age and level of degradation increased; For samples older than 2.5 billion years, the AI ​​flagged less than half as having a biotic origin, and with a lower overall confidence level.

Nonetheless, it was very old samples from South Africa that led to the team’s most dramatic findings: signs of biogenic molecules in 3.3 billion-year-old specimens from a formation called Josefsdal Chert, and evidence of ancient oxygen-producing photosynthesis in 2.5 billion-year-old rocks from the Gamohaan Formation. Pre-existing geochemical evidence meant that neither result was a surprise, but being supported by biomolecular data is a real breakthrough. “The key is that our validation set included truly unknown samples, some of which have been debated for decades,” says paper co-author Anirudh Prabhu, who studies geoinformatics at Carnegie. “And the model made independent predictions that sometimes confirmed existing suspicions.”

The most surprising discoveries came from AI foiling its human bids. The system flagged a dead shell as photosynthetic — a mistake, it seemed, until researchers realized the system had detected algae growing on the shell. A similar photosynthesis “false alarm” occurred for a wasp nest, which the AI ​​correctly linked to the chewed wood the nest was made from. “The model was good, but for the wrong reason,” says Prabhu.

Linda Kah, a geochemist at the University of Tennessee at Knoxville who was not involved in the study, calls it a “magnificent effort.” Her “big data” approach offers a road map for scientists searching for even older biosignatures, she says, and poses questions that require further investigation. For example: Do the diminishing returns of AI for the oldest and most degraded samples mean that the technique is approaching a fundamental limit of what can be recognized as biotic? Or could older samples simply contain more abiotic material because life has not yet fully infiltrated the environments available on early Earth?

The answers may come soon. The team already plans to test its AI on a larger and more diverse set of samples, including those from even deeper in Earth’s history and from a wider range of extraterrestrial sources. And some interplanetary robotic explorers, including NASA’s Curiosity rover, already carry Py-GC-MS instruments, potentially offering chances for field verification of the technique on another world.

“Studies like this bring us even closer to the origin and evolution of life on Earth,” says Amy J. Williams, a geobiologist at the University of Florida, who was also not involved in the work. “They prepare us to answer the most fundamental question of whether we are alone in the universe.”