Machine learning tool distinguishes signs of life from non-living compounds in space samples

by Clarence Oxford
Los Angeles CA (SPX) Nov 27, 2025
Researchers developed LifeTracer, a machine learning framework, to analyze mass spectrometry data from space and terrestrial samples. They used advanced two-dimensional gas chromatography and high-resolution time-of-flight mass spectrometry to study eight carbonaceous meteorites and ten terrestrial rock samples. LifeTracer applies logistic regression to compound-level features and achieved 87 percent classification accuracy, distinguishing samples derived from meteorites and Earth rocks.

In their results, scientists detected thousands of molecular peaks in each sample category - 9,475 in meteorites and 9,070 in terrestrial rocks. Key molecular differences included weight distributions and chromatographic retention times, with meteorite compounds showing greater volatility and lower retention values. These findings help define the molecular boundaries between abiotically and biotically formed materials.

Polycyclic aromatic hydrocarbons (PAHs) and their alkylated derivatives were highlighted as principal predictors within the model. Naphthalene was the most predictive compound for abiotic samples. The detection and distribution of PAHs in meteorite samples support their formation outside biological processes and help improve biosignature discrimination.

This approach goes beyond searching for specific molecular biomarkers; instead, it uses data-driven statistical analysis and computational learning to distinguish broad chemical patterns between life-related and non-life-related organics. The team noted the framework is unbiased and scalable, making it suitable for analysis of complex, uncharacterized organics expected in planetary sample-return missions. Advanced machine learning can thus improve the interpretation of ambiguous organic mixtures as future missions seek evidence of extraterrestrial life.

Research Report:Discriminating abiotic and biotic organics in meteorite and terrestrial samples using machine learning on mass spectrometry data

Related Links
NASA/Goddard Space Flight Center
Lands Beyond Beyond - extra solar planets - news and science
Life Beyond Earth