Omics Data Integration to Predict Drug Resistance in Mycobacterium tuberculosis

Tuberculosis (TB) remains the deadliest infectious disease worldwide. Improved diagnosis is essential to address increasing TB antimicrobial resistance and to assign an effective treatment regimen. The advent of whole-genome sequencing (WGS) promises to overcome slow, more expensive, and biohazardous assay susceptibility testing. However WGS relies on previous understanding of resistance mechanisms in M. tuberculosis (Mtb). To date, WGS has poor performance to predict resistance on new, lessstudied drugs like bedaquiline, failing to distinguish innocuous rare variants from mutations that confer a resistant phenotype in Mtb. We propose an innovative approach to overcome this gap by applying an integrative multi-omics analysis in Mtb to determine the aggregated effect of each mutation on drug resistance, including the contribution of rare mutations and convergent evolution and using computational tools like machine learning and Bayesian networks. Besides improving the accuracy of WGS data interpretation, this integrative analysis will allow us to explore novel cellular mechanisms involved in antimicrobial resistance, close the gap between genotypic and phenotypic drug susceptibility testing, and better understand resistance evolutionary pathways towards Mtb resistance.