Development of novel methods to predict the drug resistance phenotype of Mycobacterium tuberculosis variants.

Every year, ten million people develop tuberculosis (TB) and 1,7 million people die from TB. About 600,000 people are diagnosed with TB resistant to rifampicin, a key first-line TB drug. Drug resistant TB thus poses a global public health problem and threatens global TB control. Whole Genome Sequencing (WGS) is an innovative method to detect drug resistance, but current drug resistance tools can only predict the resistance profile for a small proportion of the over 2000 genetic variants that may be associated with resistance. Furthermore, in late 2018, the WHO will prioritize fluoroquinolones (FQ) and bedaquiline (BDQ) as two of the three core drugs for treatment of rifampicin resistant TB. While FQs are a well-studied class of antibiotics, BDQ is a new drug. Consequently, genotype-phenotype data is limited, and no mutations have been statistically associated with BDQ resistance. New tools to predict resistance are needed to make optimal clinical use of WGS. In this project, I will develop two different methods for prediction of drug resistance in Mycobacterium tuberculosis. In the first method, I will apply pattern mining methods to assess alteration of the drug binding site as a resistance mechanism, by modelling the drug-drug target interaction. In the second method, I will assess regulatory resistance mechanisms, such as upregulation of efflux pumps, by modelling transcriptional changes caused by genetic mutations.

Keywords:DRUG RESISTANCE

Disciplines:Computational biomodelling and machine learning

Project type:Collaboration project