The information content of species: formal definitions of pangenome complexity track with bacterial lifestyle

Abstract

Genes and other genomic elements have variable presence absence patterns across most bacterial species. Pangenome fluidity is often invoked to measure this genome flux. Fluid pangenomes contain genes found only in subsets of species strains. Tighter pangenomes contain more genes that define a shared core. Species definitions are often tied to this pangenome diversity. In any global comparative framework, pangenomes must be calculated across all known species. But defining pangenomes is fraught with computational and biological challenges, requiring assembly, annotation, alignment, and phylogenetics of millions of orthologs. We offer an alternate view that de-centers the gene and emphasizes the raw information content of sequences. Information is data that reduces uncertainty. Tight pangenomes, with elements repeated across every strain in a species ensemble, contain more complete information. In contrast, fluid pangenomes have more uncertainty, higher complexity, and higher information diversity. Bacterial lifestyle has been shown to drive this information diversity. For example, challenging environments often increase information diversity by encouraging the accrual of auxiliary genes. Here, we employ agile complexity metrics to quantify this increase. Ensembles of free-living, motile, and non-pathogenic species have high genomic complexity. Ensemble complexity decreases in species bound to specific hosts. Because we eliminate annotation and alignment, our method is fast enough to evaluate species across all known bacterial genomes. The approach democratizes classification and our results highlight how broad the term "species" has become.