Fig. 4: Hospital association of popPUNK-clusters and contig-clusters.

a,b, Maximum-likelihood core-gene phylogenetic tree for E. coli (a) and K. pneumoniae subsp. pneumoniae (b) showing in-hospital (dark blue), post-discharge (light blue) and community (green) isolates, where post-discharge is defined as up to 120 d post-hospital discharge. Hospital-associated samples are distributed across the tree but only the popPUNK-cluster highlighted in red shows an association with in-hospital isolation. c,d, Manhattan plots showing P value of two-sided Fisher’s exact test for association of popPUNK-cluster (c) and contig-cluster (d) with in-hospital isolation. Dotted line shows Bonferroni-corrected value corresponding to P = 0.05. Only one popPUNK-cluster is significantly associated with in-hospital isolation (highlighted in red on the plot (c) and core-gene tree (a)) at this level. Similarly, one contig-cluster is associated with in-hospital isolation, highlighted in red; this is the contig-cluster associated with the hospital-associated lineage.