Abstract EANA2025-28

The search for biosignatures—indicators of past or present life—is a fundamental challenge in astrobiology and paleobiology. Recent work by Cleaves et al. introduced a machine-learning approach to distinguishing biotic and abiotic organic matter using pyrolysis gas chromatography–mass spectrometry (Py-GC-MS) (1). My research extends these methods by decoding chemical signals embedded in spectral data to uncover molecular patterns that characterize life-derived samples, as well as potentially agnostic biosignatures—chemical fingerprints that may indicate life even in the absence of Earth-like biology.

To achieve this, I analyzed spectral data from a wide array of organic materials, including modern biological samples, fossilized remains, carbonaceous meteorites, and laboratory-synthesized compounds. This integrated dataset bridges prebiotic chemistry, early evolution on Earth, and extraterrestrial organics. While previous studies have demonstrated that machine learning can distinguish between recent and taphonomically altered biogenic materials, significant ambiguity remains—particularly in differentiating biotic from abiotic origins. My work addresses this challenge by incorporating additional molecular-level detail extracted from spectral data, improving our ability to differentiate organic materials by origin and formation process.

This research has a dual impact. First, it enhances our ability to interpret the earliest traces of life on Earth by providing a refined chemical lens through which to view ancient organic matter. Second, it directly informs the search for life beyond Earth: several Mars rover missions (including Viking and Curiosity) and the upcoming Titan Dragonfly mission are equipped with or plan to use Py-GC-MS instruments. By characterizing spectral features of organic compounds in both terrestrial and extraterrestrial contexts, my work assists in laying the groundwork for identifying biosignatures in planetary exploration.

Ultimately, this research contributes to the development of agnostic, data-driven biosignature frameworks, capable of detecting life-related chemistry without relying on assumptions of Earth-like biochemistry. It advances our capacity to recognize biological function across time and space, even in the absence of morphology or context—transforming how we search for life in the universe and understand its earliest emergence on Earth.

 

[1] Cleaves, H. James, et al. "A robust, agnostic molecular biosignature based on machine learning." Proceedings of the National Academy of Sciences 120.41 (2023): e2307149120.