A robust, agnostic molecular biosignature based on machine learning

Abstract

The search for definitive biosignatures-unambiguous markers of past or present life-is a central goal of paleobiology and astrobiology. We used pyrolysis-gas chromatography coupled to mass spectrometry to analyze chemically disparate samples, including living cells, geologically processed fossil organic material, carbon-rich meteorites, and laboratory-synthesized organic compounds and mixtures. Data from each sample were employed as training and test subsets for machine-learning methods, which resulted in a model that can identify the biogenicity of both contemporary and ancient geologically processed samples with ~90% accuracy. These machine-learning methods do not rely on precise compound identification: Rather, the relational aspects of chromatographic and mass peaks provide the needed information, which underscores this method's utility for detecting alien biology.

Keywords: biosignatures; carbonaceous meteorites; machine learning; organic chemistry; taphonomy.

Fig. 1.

Combined three-dimensional Pyr-GC-EI-MS data for complex organic mixtures in carbonaceous meteorites (A) and microbial samples (B). These graphs display peak intensities (vertical scale, normalized to the highest peak intensity) for 3,000 elution time bins (right-hand scale) and their mass spectra over 150 m/z bins (left-hand scale). Green circles with vertical “stems” do not represent intensity values, but rather features the machine-learning algorithm recognizes as important discriminants among samples.

Fig. 2.

Grouping of samples according to the machine-learning methods explored here. Biologically derived samples (green/blue) are distinguished from abiotic samples (orange). Taphonomically altered biological samples (blue) lie along a trend distinct from that of contemporary biological samples (green).