[This corrects the article DOI: 10.1371/journal.pcbi.1007732.].

MOTIVATION: Horizontal gene transfer (HGT) is a major source of variability in prokaryotic genomes. Regions of genome plasticity (RGPs) are clusters of genes located in highly variable genomic regions. Most of them arise from HGT and correspond to genomic islands (GIs). The study of those regions at the species level has become increasingly difficult with the data deluge of genomes. To date, no methods are available to identify GIs using hundreds of genomes to explore their diversity. RESULTS: We present here the panRGP method that predicts RGPs using pangenome graphs made of all available genomes for a given species. It allows the study of thousands of genomes in order to access the diversity of RGPs and to predict spots of insertions. It gave the best predictions when benchmarked along other GI detection tools against a reference dataset. In addition, we illustrated its use on metagenome assembled genomes by redefining the borders of the leuX tRNA hotspot, a well-studied spot of insertion in Escherichia coli. panRPG is a scalable and reliable tool to predict GIs and spots making it an ideal approach for large comparative studies. AVAILABILITY AND IMPLEMENTATION: The methods presented in the current work are available through the following software: https://github.com/labgem/PPanGGOLiN. Detailed results and scripts to compute the benchmark metrics are available at https://github.com/axbazin/panrgp_supdata.

The use of comparative genomics for functional, evolutionary, and epidemiological studies requires methods to classify gene families in terms of occurrence in a given species. These methods usually lack multivariate statistical models to infer the partitions and the optimal number of classes and don't account for genome organization. We introduce a graph structure to model pangenomes in which nodes represent gene families and edges represent genomic neighborhood. Our method, named PPanGGOLiN, partitions nodes using an Expectation-Maximization algorithm based on multivariate Bernoulli Mixture Model coupled with a Markov Random Field. This approach takes into account the topology of the graph and the presence/absence of genes in pangenomes to classify gene families into persistent, cloud, and one or several shell partitions. By analyzing the partitioned pangenome graphs of isolate genomes from 439 species and metagenome-assembled genomes from 78 species, we demonstrate that our method is effective in estimating the persistent genome. Interestingly, it shows that the shell genome is a key element to understand genome dynamics, presumably because it reflects how genes present at intermediate frequencies drive adaptation of species, and its proportion in genomes is independent of genome size. The graph-based approach proposed by PPanGGOLiN is useful to depict the overall genomic diversity of thousands of strains in a compact structure and provides an effective basis for very large scale comparative genomics. The software is freely available at https://github.com/labgem/PPanGGOLiN.