Bill Hanage: Lineage calling using short k-mers can identify drug resistant clones in minutes

Bill Hanage of Harvard T. H. Chan School of Public Health presented a novel method for extremely rapid detection and identification of antimicrobial resistant (AMR) bacteria from clinical research samples. Bill kicked off his presentation by highlighting the impact of drug resistant infections, which in the US alone caused over 2 million illnesses and 23,000 deaths. Startingly, the number of deaths attributable to AMR is anticipated to rise to 10 million by 2050, surpassing the expected number of deaths from cancer. Bill also drew attention to carbapenem-resistant Enterobacteriaceae (CRE), which have become resistant to all or nearly all available antibiotics. Clearly, rapid identification of AMR is paramount to treatment and control of these threats to human health.

According to Bill, for almost all pathogens, drug resistance can be associated with a specific lineage. He pointed out that clinical microbiologists have been reporting likely susceptibilities for many years. Based on this, he posed the question: ‘why are we so busy trying to detect the gene, when we could be trying to detect the clone’. The latter scenario can be achieved much faster and easier – especially in combination with real-time nanopore sequencing.

To test this concept, the team at Harvard T. H. Chan School of Public Health, created RASE, which stands for Resistance Associated Sequence Element. The essence of this workflow is the development of a database of high-quality reference sequence k-mers for susceptible and resistant bacteria. The nanopore sequencing reads are matched against the k-mers in real time to identify resistance based on identity with known bacterial lineages. Bill explained that k-mers of 18 nucleotides were used as this length provides discriminatory alignment and reduces the likelihood of sequencing errors that could prevent matching.

Using RASE, the team were able to identify susceptible and resistant isolates from metagenomic samples in just 5 minutes. Bill also pointed out that the platform is very memory efficient, allowing analysis on a laptop, which, when combined with the portable MinION device, is ideal for potential clinical and outbreak settings.

Underlining the concept behind this novel approach, Bill closed his presentation by showing an image of a tiger and suggesting that: ‘if you were walking in a forest and you heard a growl and saw something large and stripy, you wouldn’t wait to sequence it to see if it had claws’.