Use of multivariate statistics for 16S rRNA gene analysis of microbial communities

Abstract

Understanding dynamic processes and diversity in microbial communities is of key importance for combating pathogens and for stimulating beneficial bacteria. We have addressed these challenges utilising multivariate statistics for analyses of microbial community structures. We based our microbial community analyses on 16S rRNA gene data. This gene is by far the most widely applied genetic marker for phylogenetic and microbial community studies. Both probe and clone library data were analysed. We analysed the clone library data using a newly developed coordinate-based phylogenetic approach. By using coordinates, we avoid both DNA sequence alignments and the need for definition of operational taxonomic units (OTUs). The basic principle is to transform the sequence data to frequencies of multimers (short sequences of n = 2 to 6), and then to use principal component analyses (PCA) for data compression into an orthogonal coordinate space. We used our coordinate method for global 16S rRNA gene analyses of prokaryotes. When comparing microbial communities, it is often important to determine the relationship between the microflora and knowledge about the samples analysed. We used partial least square regression (PLSR) to relate physical/chemical properties to microbial community composition. This was done by analysing both probe and clone library data using the effect of modified atmosphere packaging (MAP) on fish microflora as an example. We are currently investigating approaches to describe dynamic microbial community interactions. Our ultimate goal is to understand and model the main dynamic interactions in complete microbial communities.