Fig. 5: Deciphering the KL type of the infected ancestral lysogens.

A Outline of the algorithm used to determine the infected ancestor of each prophage to infer the KL type at the time of infection. Three scenarios are illustrated using phylogenetic trees to showcase how the methodology functions under different conditions, with KL types represented by blue and green colors. B In a straightforward scenario, the prophage is part of a vOTU identified in strains within the same clade (e.g., strains D3 and D4) but absent from other clades at the next hierarchical level (n + 1) connected to node B. Here, the KL type at the time of infection is inferred from the last common ancestor of these strains, corresponding to node C2. C In this scenario, the prophage's presence in strains D1, D3, and D4 suggests a single infection event. Consequently, the KL type at the time of infection aligns with that of the ancestor at node B. D In cases where the phylogenetic tree lacks adequate support (e.g., a missing branch connecting leaf D5 to node A), the inferred infected ancestor would be node B, despite the actual infection event occurring earlier. This would lead to an incorrect inference of the prophage's specificity as blue, due to the absence of additional information. E Schematic representation of the pipeline used to create a labeled collection of prophages from the Klebsiella genome dataset. Representations of phages and bacteria are from BioRender.