Biosignatures & Paleobiology
Status Report
Earth and Space Science
March 25, 2025
Interpretable Machine Learning Biosignature Detection From Ocean Worlds Analogue CO2 Isotopologue Data
(a) Visualization of statistical effects of the highly predictive biosignature features using a Regression-based Association-Interaction Network (RAIN) with node numbers in order of NPDR-LURF importance. RAIN encodes feature main effects and statistical interactions in a network. Nodes 2, 3, 4, and 5 have the largest main effects. For nodes 2 and 4, the main effect increases the probability of an abiotic class prediction. The TS feature diff2_acff1 is also a hub in the interaction network, participating in three large interactions with nodes 2, 3, and 5 that influence the probability for both biotic and abiotic predictions. (b) The Random Forest classifier trained on NPDR-LURF features for biosignatures yields a training accuracy of 87.9%, trained on a sample size m = 140. Dark gray and green diagonals represent the number of correct predictions: 81/89 abiotic samples are correctly classified, and 42/51 biotic samples are correctly classified. The precision for biotic samples in the training data is 0.824 and the recall is 0.84. (c) The same classifier yields a high test accuracy of 88.2%: 20/22 abiotic samples are correctly classified, and 10/12 biotic samples are correctly classified. The biotic precision for the test data is 0.833 and the recall is also 0.833. β Earth and Space Science
Future missions to icy ocean worlds (OW) such as Europa and Enceladus will evaluate the habitability and potential for biosignatures on these worlds.
These missions will benefit from autonomous science and machine learning (ML) methods to process high volumes of data and prioritize signals of interest for the first available downlink. Mass spectrometers (MS) are suitable instruments for implementing science autonomy due to their rich spectral data products and potential for biosignature detection.
Light stable isotopes are strong candidates for biosignatures due to the large fractionations promoted by biological activity. However, complex abiotic geochemistry may obscure or mimic biogenic isotope fractionations. ML may accurately disentangle biosignatures from abiotic mimicry in MS data; however, ML model predictions can be inscrutable to human interpretation, compromising trust in scientifically significant detections.
We develop and test a new biosignature detection ML model using a novel, laboratory-generated, CO2 isotopologue data set of analogue OW samples. These data include various potential OW seawater chemistries and biotic mimicry. Our ML approach includes feature (variable) construction, providing mathematical and geochemical context for biosignatures, and a feature selection method called Nearest-neighbors Projected Distance Regression (NPDR) that identifies important predictors.
Our Random Forest biosignature model predicts the presence of biosignatures with 87.3% mean accuracy regardless of the sample brine chemistry. We add network visualization of main effects and statistical interactions for interpretation of model prediction mechanisms.
We use single-sample (local) variable importance scores to diagnose false predictions for individual samples, which is crucial for trust in astrobiology ML biosignature models.
Interpretable Machine Learning Biosignature Detection From Ocean Worlds Analogue CO2 Isotopologue Data, Earth and Space Science (open access)
Astrobiology
Explorers Club Fellow, ex-NASA Space Station Payload manager/space biologist, Away Teams, Journalist, Lapsed climber, Synaesthete, NaβVi-Jedi-Freman-Buddhist-mix, ASL, Devon Island and Everest Base Camp veteran, (he/him) ππ»