Identifying disease-associated dysbiosis in image-based spatial microbiome via MiSA
En palabras de los autores
Background: Extensive studies have revealed the associations of microbiome dysbiosis with disease through compositional analysis of 16S rRNA amplicon or shotgun metagenomic sequencing data. However, dysbiosis in spatial organization of microbiome inhabiting the body sites and its implication for diseases remain understudied. Recently developed high phylogenetic resolution fluorescence in situ hybridization (HiPR-FISH) is an image-based mapping technology that provides spatial microbiome data at single-cell resolution, which allows for visualizing and quantifying spatial arrangement of microbial cells. Despite the growing availability of such data, no methods have been tailored to studying spatial dysbiosis in image-based microbiome data. Results: Here we introduce MiSA (Microbiome Spatial Aberration), the first statistical framework for identifying disease-associated spatial dysbiosis and the individual taxa that may contribute to it, based on differing degrees of clustering or regularity, i.e., microbial cells being more aggregated or more evenly spaced than expected under complete spatial randomness, between diagnostic groups. In simulations based on a real dataset, MiSA was robust across diverse and noisy spatial configurations, achieving substantially higher statistical power than a method borrowed from spatial proteomics, while controlling type I error at both the individual-taxon and community levels. When applied to an oral biofilm study, MiSA detected peri-implant mucositis-associated spatial dysbiosis at the community level and identified five taxa as potential contributors, including Actinomyces, which showed increased clustering in mucositis. Conclusions: MiSA is a powerful and robust framework for identifying spatial dysbiosis and the contributing taxa in image-based microbiome data, offering new insights into the spatial organization of microbial communities in disease.
Apareció: jueves, 19 de noviembre. bioRxiv. Preprint, todavía sin revisión por pares.