7-5 Decoding Microbial Life at Species Resolution in Earth’s Dynamic Environments

Session: Earth Life Sciences across the Cordillera (Posters)

Presenting Author:

Authors:

Abstract:

Microbial life plays a fundamental role in shaping Earth’s systems, mediating key biogeochemical cycles of carbon, nitrogen, iron, and other elements that link biological activity with geological and marine processes. In tectonically active and environmentally heterogeneous regions, such as those influenced by volcanism, rifting, and marine–terrestrial interactions, microbial communities contribute to mineral transformations, element cycling, and ecosystem resilience. These organisms inhabit diverse environments, including marine sediments, hydrothermal systems, and arid landscapes, offering critical insights into the co-evolution of life and Earth’s dynamic geologic history.

Accurate identification and taxonomic resolution of microbial communities are therefore essential for geogenomic studies aimed at understanding biological responses to geological processes and environmental change. Traditional microbial identification methods are often slow, rely on cultivation of isolated strains, and fail to capture the vast uncultivable diversity present in natural systems. While metagenomics enables the recovery of genomic information directly from environmental samples without cultivation, it generates millions of sequencing reads that require specialized computational tools and substantial resources for analysis. Moreover, despite recent advances, taxonomic assignment of metagenomic data remains challenging, particularly at the species level.

Recent developments in deep learning have shown promising results for taxonomic classification of metagenomic sequences; however, achieving reliable species-level resolution remains an open problem. In this work, we propose a deep learning–based methodology for species-level taxonomic assignment from metagenomic datasets, designed specifically to address the complexity and scale of environmental genomic data. Our approach employs multiple genus-specific models in a modular framework that improves both accuracy and computational efficiency compared to single, global classifiers. By refining taxonomic resolution, this methodology enhances our ability to characterize microbial diversity in geologically complex environments and supports geogenomic investigations of microbial evolution, adaptation, and ecosystem function across Earth’s dynamic landscapes.

Geological Society of America Abstracts with Programs. Vol. 58, No. 3, 2026

© Copyright 2026 The Geological Society of America (GSA), all rights reserved.