From microbial diversity to functional potential using dimensionality reduction

Abstract

Machine Learning (ML) has become an increasingly prevalent tool in microbial oceanography. The high dimensionality of microbial diversity data from 'omics observations is highly suitable for ML analysis, with many recent studies showcasing its utility for exploratory ecological feature finding and process prediction. Here, we compare the Self Organizing Map (SOM) to two other well-documented dimensionality reduction methods including Principal Coordinate Analysis (PCoA) and Weighted Gene Correlation Network Analysis (WGCNA) using near daily 16S rRNA gene amplicon sequencing data from the 2019-2020 MOSAiC International Arctic Drift Expedition. We then compare k-means clustering outputs from each method to available metagenomes, extracting functionally distinct seasonal microbial ecotypes in the surface Arctic Ocean. Our results indicate the SOM method better represented expected seasonal transitions and identified a greater number of metabolically distinct functional groups than the more traditional PCoA ordination. Investigating the importance of biological context in dimensionality reduction, we also compare these sample-based functional outputs to a taxa clustering approach using a k-means adapted WGCNA correlation network. Ultimately, we identified 4 community ecotypes with distinct taxonomic and functional cut-offs driven by seasonality, water mass, and substrate turnover, highlighting the importance of succession in functional diversity for the central Arctic Ocean. These results further reinforce ML methodologies as a meaningful translator in the mining of historical amplicon datasets to address modern mechanistic questions and potentially provide 'omics informed ecotype diversity to leverage in mechanistic biogeochemical models.