Unravelling the co-occurrence patterns of antimicrobial resistance within 214K metagenomes
The indirect selection of antimicrobial resistance genes (ARGs) of different resistance classes can have damaging effects during antimicrobial treatment. However, understanding the co-occurrence of ARGs in microbiomes is a challenge, which becomes even more complex when working with metagenomes as the true abundance remains unknown. With the large data collection of read count results stemming from the alignment of 442 Tbp of sequencing reads from 214K metagenomic samples to ARGs, we have investigated the correlation of ARG abundances across the whole data collection and in selected sampling sources.
Using the whole data collection, we observed 225 ARGs of different resistance classes correlating with each other through 2,344 correlation edges (p-value ≤ 0.01, correlation ≥ 0.6). We saw multiple instances of ARGs co-occurring despite conferring resistance to different antibiotic classes. For example, genes giving resistance to beta-lactams often had many connections to ARGs for aminoglycoside and quinolone resistance. However, these correlations were mainly found in human and livestock samples, not soil and water samples.
These findings demonstrate that there can be a heightened risk of using one class of antibiotics can increase the abundance of other types of resistances through indirect selection, which can be used to create guides on how to responsible use antimicrobials in different environments.
Link to preprint on the 214K metagenomic data collection:
https://www.biorxiv.org/content/10.1101/2022.05.06.490940v1
Hannah-Marie Martiny's presentation