Astrobiology in the Era of Big Data: Leveraging Machine Learning for Prebiotic Chemistry and Planetary Exploration

Searching for evidence of life on other worlds and elucidating conditions conducive for origins of life events are two of the 12 priority science questions to be addressed by NASA in the coming decade (2022-2032 Planetary Science & Astrobiology Decadal Survey). Prebiotic chemistry can inform the search for life since abiotic mechanisms can generate molecules that could be mistakenly identified as biosignatures (Barge et al., 2022). Conversely, astrobiology missions can inform origins of life studies by serving as a means of ground truthing experimental research and providing an avenue for exploring changes in organic trends over geologic time scales. For both origin of life and life detection, contextualizing organic content with its geochemical environment is key. However, planetary samples and prebiotic mixtures that simulate the natural environment are complex, obscuring analyses done in the lab or in-situ. While Earth-based analyses allows samples to be characterized with gold standard instruments, the resulting datasets can be large and difficult to parse, especially with regards to organics. For ocean world missions, the issue of data complexity is compounded by constraints on data volumes that can be downlinked to Earth and the communication delay which hinders ground-based mission operations. Given these constraints and the limited lifetime of such a mission, autonomous exploration is mission-critical, especially for any life detection mission where multiple instruments would be used to provide environmental context for putative prebiotic organics or biosignatures.

In this talk, I will highlight efforts to develop machine learning (ML) strategies to characterize complex spectral data sets acquired via mission-relevant techniques namely, Laser Induced Breakdown Spectroscopy, Raman spectroscopy, and High-Resolution Mass Spectrometry (HRMS). I will review results that demonstrate the utility of ML to: classify organic compounds via tandem-HRMS and decipher organic and geochemical trends in hydrothermal vent precipitates and prebiotic mixtures which serve as analogs for ocean world geochemical processes. I will discuss the potential for these strategies to be leveraged for the field of prebiotic chemistry, the analysis of returned samples, and the autonomous exploration of life-detection missions to ocean worlds.

Barge LM, Rodriguez LE, Weber JM, Theiling BP. Determining the ‘‘Biosignature Threshold” for Life Detection on Biotic, Abiotic, or Prebiotic Worlds. Astrobiology 22, 4, 2022.