Classifying mobile genetic elements and their interactions from sequence data: The importance of existing biological knowledge.

We agree with Che et al. (1) that understanding how mobile genetic elements (MGEs) spread antimicrobial resistance (AMR) is important. However, decades of research have already characterized a diverse toolbox of MGEs involved in the emergence of AMR, including conjugative plasmids and insertion sequences (ISs), and their interactions. “Mobile” AMR generally arises following rare capture of a chromosomal gene by a particular “intracellularly mobile” MGE, e.g., an IS or a transposon (Tn), and translocation to “intercellularly mobile” MGEs [plasmids, phage, integrative elements (2, 3)]. Tools for systematic analysis of AMR gene−MGE interactions in mounting sequence data are needed, but incorporating existing biological knowledge into their design is vital, as are representative benchmarking datasets and recognition of data biases and resulting limitations. Che et al. screened bacterial plasmid sequences for markers to classify them as nonmobilizable (no relaxase), mobilizable (relaxase), or conjugative (relaxase+). Relying on the absence of a predicted feature is risky, and Plascad’s poor sensitivity in many species was not recognized, due to a taxonomically inadequate benchmark dataset. Numerous conjugative plasmids, and close relatives, from low-GC gram-positive organisms (dataset S1 in ref. 1; e.g., Staphylococcus, pSK41, pWBG4, pWBG749; Clostridium, pCW3, pCP13; Enterococcus/Streptococcus , pAMβ1, pCF10, pRE25, pMG1, pIP501) (3⇓–5 … [↵][1]1To whom correspondence may be addressed. Email: sally.partridge{at}health.nsw.gov.au or neville.firth{at}sydney.edu.au. [1]: #xref-corresp-1-1