A robust, agnostic molecular biosignature based on machine learning.

Area of Science:

• Paleobiology
• Astrobiology
• Analytical Chemistry

Background:

• The identification of definitive biosignatures is crucial for understanding past and present life on Earth and potentially elsewhere.
• Analyzing complex organic samples requires advanced techniques to distinguish biological from non-biological origins.

Purpose of the Study:

• To develop and validate a machine-learning model for accurately identifying biogenic materials.
• To assess the model's effectiveness across diverse sample types, including living cells, fossils, meteorites, and synthetic compounds.

Main Methods:

• Pyrolysis-gas chromatography coupled to mass spectrometry (Py-GC-MS) was used to analyze various organic samples.
• Machine-learning algorithms were trained on data from chemically disparate samples to recognize patterns indicative of life.

Main Results:

• The developed machine-learning model achieved approximately 90% accuracy in identifying the biogenicity of both contemporary and geologically processed samples.
• The model's success relies on relational data from chromatographic and mass peaks, not just specific compound identification.

Conclusions:

• This machine-learning approach offers a robust method for biosignature detection, independent of precise molecular identification.
• The technique holds significant promise for the search for alien biology in astrobiological investigations.