P52

Resistance to the first-line drugs isoniazid (INH) and rifampicin (RIF) in Mycobacterium tuberculosis (multidrug-resistant TB; MDR-TB) threatens disease control. Mutations that confer resistance to anti-TB drugs often come with a fitness cost. To overcome these fitness costs, M. tuberculosis can develop further, beneficial, compensatory mutations in genes that code for proteins performing similar biological roles or that interact with drug-targets. Here we leverage the presence of known and putative compensatory mutations along with phenotypic and genotypic data to detect the presence of novel mutations occurring in the target genes of INH (katG) and RIF (rpoB), inferring their causative role in resistance to these drugs. Across ~32k isolates, there were 6,262 (19.17%) with INH and 5,435 (16.64%) with RIF phenotypic resistance. Of these, known mutations in katG and rpoB explained most of the resistance, however ~25k (78%) and ~25k (78%) samples did not have any known resistance mutations in INH or RIF target genes, respectively. By examining isolates with compensatory mutations but no known resistance mutation, we identified a number of inferred novel resistance mutations. We also characterise the putative katG mutations in terms of their co-occurrence with other INH variants - katG-Ser315Thr and other INH genes, e.g. fabG1. Detecting resistance mutations from whole genome sequencing can inform clinical management, but also revealed new variants linked to resistance. We have identified new resistance markers and, after validation, have added to a list of mutations for genotypic drug-resistance predictions.