First shotgun metagenomics study of Juan de Fuca deep-sea sediments reveals distinct microbial communities above, within, between, and below sulfate methane transition zones

The marine deep subsurface is home to a vast microbial ecosystem, affecting biogeochemical cycles on a global scale. One of the better-studied deep biospheres is the Juan de Fuca (JdF) Ridge, where hydrothermal fluid introduces oxidants into the sediment from below, resulting in two sulfate methane transition zones (SMTZs). In this study, we present the first shotgun metagenomics study of unamplified DNA from sediment samples from different depths in this stratified environment. Bioinformatic analyses showed a shift from a heterotrophic, Chloroflexota-dominated community above the upper SMTZ to a chemolithoautotrophic Proteobacteria-dominated community below the secondary SMTZ. The reintroduction of sulfate likely enables respiration and boosts active cells that oxidize acetate, iron, and complex carbohydrates to degrade dead biomass in this low-abundance, low-diversity environment. In addition, analyses showed many proteins of unknown function as well as novel metagenome-assembled genomes (MAGs). The study provides new insights into microbial communities in this habitat, enabled by an improved DNA extraction protocol that allows a less biased view of taxonomic composition and metabolic activities, as well as uncovering novel taxa. Our approach presents the first successful attempt at unamplified shotgun sequencing samples from beyond 50 meters below the seafloor and opens new ways for capturing the true diversity and functional potential of deep-sea sediments.

Generalized radial basis function networks for classification and novelty detection: self-organization of optimal Bayesian decision

By adding reverse connections from the output layer to the central layer it is shown how a generalized radial basis functions (GRBF) network can self-organize to form a Bayesian classifier, which is also capable of novelty detection. For this purpose, three stochastic sequential learning rules are introduced from biological considerations which pertain to the centers, the shapes, and the widths of the receptive fields of the neurons and allow ajoint optimization of all network parameters. The rules are shown to generate maximum-likelihood estimates of the class-conditional probability density functions of labeled data in terms of multivariate normal mixtures. Upon combination with a hierarchy of deterministic annealing procedures, which implement a multiple-scale approach, the learning process can avoid the convergence problems hampering conventional expectation-maximization algorithms. Using an example from the field of speech recognition, the stages of the learning process and the capabilities of the self-organizing GRBF classifier are illustrated.