P-DOR, an easy-to-use pipeline to reconstruct bacterial outbreaks using genomics

Abstract

Summary: Bacterial Healthcare-Associated Infections (HAIs) are a major threat worldwide, which can be counteracted by establishing effective infection control measures, guided by constant surveillance and timely epidemiological investigations. Genomics is crucial in modern epidemiology but lacks standard methods and user-friendly software, accessible to users without a strong bioinformatics proficiency. To overcome these issues we developed P-DOR, a novel tool for rapid bacterial outbreak characterization. P-DOR accepts genome assemblies as input, it automatically selects a background of publicly available genomes using k-mer distances and adds it to the analysis dataset before inferring a Single-Nucleotide Polymorphism (SNP)-based phylogeny. Epidemiological clusters are identified considering the phylogenetic tree topology and SNP distances. By analyzing the SNP-distance distribution, the user can gauge the correct threshold. Patient metadata can be inputted as well, to provide a spatio-temporal representation of the outbreak. The entire pipeline is fast and scalable and can be also run on low-end computers.

Availability and implementation: P-DOR is implemented in Python3 and R and can be installed using conda environments. It is available from GitHub https://github.com/SteMIDIfactory/P-DOR under the GPL-3.0 license.

Figure 1.

The P-DOR outputs of the test analysis. (A) Phylogenetic tree of the Analysis Dataset (AD). The first column shows epidemiological clusters of strains with SNP distances below the threshold set by the user. In addition, a heatmap representing the detection of resistance and virulence determinants is shown next to the tree for a better representation of the epidemic event. (B) Distribution of SNP distances calculated between all permutations of genome pairs in the AD. (C) Timeline depicting the movements of the patients during the hospitalization. Points indicate outbreak genomes and are shaped according to the isolation source of the corresponding strain. Samples are linked if their genetic distance in terms of SNPs is below the threshold.