New preprint: automatically building a better bedaquiline catalogue

A catalogue recording whether individual mutations confer resistance or not to specified antibiotics is a necessary component of genetics-based clinical microbiology. Such catalogues need to be not only accurate but also meet a number of minimum requirements if they are to be used widely. Dylan Adlard, who is studying for his PhD, has developed a python package, catomatic, that automatically applies some logic evolved from the approach put forward in 2015. The beauty of this is that, given a suitable set of data tables containing phenotypic drug susceptibility and genetic data, one can build a resistance catalogue based around some chosen input parameters in a few minutes. What’s more the catalogue is output in the form that piezo can understand so it can be hot-swapped in a matter of minutes.

Bedaquline (BDQ) binds to and therefore inhibits the spinning of the rotor of ATPase in M. tuberculosis leading to cell death. It is a key component of the new BPaLM all-oral regimen that the WHO has recently recommended for treating multi-drug resistant tuberculosis, yet resistance has been observed in many different countries. BDQ was included in the second edition of the WHO resistance catalogue that was released in November 2023 and here we use catomatic to reproducibly and rapidly build a BDQ resistance catalogue with slightly improved performance. As additional data becomes available being able to rapidly build new catalogues for use in genetics-based clinical microbiology (for example, EIT Pathogena) will be an important capability in our response to the rise in BDQ resistance.

His preprint is here.