Explore how machine learning is transforming the quest for ancient life, unlocking groundbreaking discoveries and insights.
The precise origins of life on Earth, a planet that emerged around 4.5 billion years ago, remain a profound mystery. Hypothetically, life may have sprouted from microorganisms in hydrothermal ocean vents several hundred million years later. To uncover the earliest forms of life, scientists often rely on the study of fossils. Notably, 3.5 billion-year-old stromatolite samples from Australia, known as microbial deposits, represent some of the oldest evidence of life. Parallel findings exist in South Africa, with similar aged microbial structures.
Another intriguing method to explore life’s history involves seeking chemical signatures of organisms on ancient rocks. However, discerning whether these signatures are due to biological activity or natural geological processes proves challenging. Furthermore, extending our understanding of these traces to distant past epochs has been difficult. Nonetheless, researchers at the Carnegie Science Institute have developed a groundbreaking machine learning algorithm to address these challenges. Initially trained to distinguish between biological and non-biological organic molecules, this tool achieved an impressive 90% accuracy rate. As emphasized by co-author Robert Hazen, this innovation allows the detection of life traces too minute for human perception.
In their study, rock samples analyzed dated back up to 3.8 billion years. Remarkably, the oldest identified sample bearing life traces was the 3.3-billion-year-old Josefsdal Chert from South Africa. Hazen remarks on the ability to uncover these ‘chemical echoes’ of ancient life beyond mere fossils. Importantly, this doesn’t imply older rocks lack biological residues; it is believed that such traces may degrade beyond detection over time. The study’s significance lies in extending the timeline for detecting chemical signatures from 1.6 billion to 3.3 billion years ago.
Moreover, the team uncovered a crucial revelation regarding the history of photosynthesis. Through their analysis, evidence of photosynthesis was found in 2.52 to 2.3 billion-year-old rocks from South Africa and Canada, indicating its inception 800 million years earlier than previously assumed. Anirudh Prabhu, another co-author, stresses that this biosignature method can differentiate not just living from non-living substances, but also between different life forms, like photosynthetic organisms.
Significantly, this advancement could aid in the quest for ancient life beyond Earth, specifically on Mars. The method is poised for use on extraterrestrial samples from the Red Planet and potentially the moons of Saturn and Jupiter, locations considered viable for life. Hazen enthuses about the study, stating, “This work marks a major leap in deciphering the world’s oldest biological traces.” Through a synergy of advanced chemical analysis and machine learning, scientists have found a way to interpret the molecular ‘ghosts,’ revealing secrets of early life from billions of years ago.
