Contingency, Repeatability and Predictability in the Evolution of a Prokaryotic Pangenome

Abstract

Pangenomes exhibit remarkable variability in many prokaryotic species. This variation is maintained through the processes of horizontal gene transfer and gene loss. Repeated acquisitions of near-identical homologs can easily be observed across pangenomes, leading to the question of whether these parallel events potentiate similar evolutionary trajectories, or whether the remarkably different genetic background of the recipients mean that post-acquisition evolutionary trajectories end up being quite different. In this study, we present a machine learning method that predicts the presence or absence of genes in individual genomes of a 2,500 genome sample of the Escherichia coli pangenome, based on the presence or absence of other accessory genes within the same genome. We are, in effect, asking whether gene acquisitions potentiate similar evolutionary trajectories or not. We found that the presence or absence of a substantial set of genes is highly predictable, from other genes alone, indicating that natural selection potentiates and maintains gene-gene co-occurrence and avoidance relationships deterministically over long-term bacterial evolution, despite differences in host evolutionary history. We propose that the pangenome can be understood as a set of genes with relationships that govern their likely cohabitants, analogous to an ecosystem's set of interacting organisms. Our findings highlight intra-genomic gene fitness effects as key drivers of prokaryotic evolution, with ensuing pangenome-wide emergence of repeated patterns of community structure.

Intra-genomic genes-to-genes correlation enables genome representation

“…According to previous studies 10–12 , certain bacterial genes had significant co-occurrence/avoidance patterns. Thus, the correlation between accessory gene status and cgST in Supplementary Fig 6 might actually result from the gene co-occurrence phenomenon.…”

“…Regarding to relationship among genes, previous studies 10–12 had found that large amount of bacterial genes could display co-occurrence or avoidance patterns with one another, and some studies on Listeria monocytogenes had presented a strong correlation between core genome based sublineage and accessory gene presence 13–16 . In L. monocytogenes , internalin A ( InlA ) is an example of gene-gene sequence variants correlation.…”

Comparative genomics reveals distinct diversification patterns among LysR-type transcriptional regulators in the ESKAPE pathogen Pseudomonas aeruginosa

“…In this regard, one must also consider the emerging role of intra-genomic selective pressures in determining the composition of prokaryotic genomes and that these niche-adapted genes may have positive or negative effects on other genes within the pangenome [102,103]. The absence of LTTR encoding genes from a 608 kb region of the P. aeruginosa genome will be fascinating to investigate for such patterns of exclusion and inclusion.…”

Assessing computational predictions of antimicrobial resistance phenotypes from microbial genomes

“…This indicates the importance of comprehensive training data. While evolution can be both contingent and deterministic [88]; a ML-based investigation [89] aligns with the deterministic theory, and some methods [42,90] have achieved moderate AMR phenotyping accuracy using phylogenetic information, indicating potential efficacy of exclusively utilizing ancestral lineages for training ML models in the absence of comprehensive data. Lastly, the third hypothesis suggests that the combined effects of AMR variants and lineage backgrounds could impact the efficacy of ML-derived predictors, as indicated by the genome misclassification frequency.…”