top of page


Whole genome sequencing for tuberculosis: it works, how do we get it used more widely?

A E Brankin(1) D Adlard(1) J Westhead(1) S Valley Omar(2) P W Fowler(1)

1:University of Oxford; 2:National Institute for Communicable Diseases

Whole genome sequencing is becoming a key method for inferring which drugs can be used to treat tuberculosis infections. To reduce sequencing errors, bioinformatic tools are usually setup so majority of reads are required to call a genetic variant; one unfortunate consequence of this approach is that minor populations are not usually detected.

By analysing 8,138 samples from the CRyPTIC project we show how allowing just two reads supporting A90V or D94G in gyrA increases the sensitivity of moxifloxacin resistance prediction from 85.4% to 94.0% with no significant change in specificity (similar result for levofloxacin). Interestingly the proportion of the reads containing the resistant variant does not correlate with the MIC [1]. These results are important since the fluoroquinolones are increasingly being included in treatment regimes yet, despite it being generally accepted that we understand their resistance mechanism, their sensitivities have remained lower than those of first-line drugs like rifampicin. 

Such observations are, however, useless unless they can be put into practice: we will therefore describe how such rules can be incorporated into resistance catalogues parseable by piezo. This software can then included in a larger bioinformatic workflow, gnomonicus, which consumes a VCF file and a catalogue and returns the antibiogram to the user. 

Finally, we will examine evidence from the CRyPTIC dataset that minor populations play a role in conferring resistance to other drugs and share some preliminary data examining the impact of minor populations on bedaquline resistance. 

1. Brankin AE, Fowler PW (2023). JAC-Antimicrobial Resistance 5:dlad039


bottom of page