geno2phenoTB
geno2phenoTB is a machine learning based tool to predict resistance of Mycobacterium tuberculosis against antibiotics using whole-genome sequencing data.
Approximately 182.000 (range 113.000-250.000) patients die each year from tuberculosis (TB) caused by a multidrug-resistant (MDR) or rifampicin-resistant (RR) strain of the Mycobacterium tuberculosis complex (MTBC). The design of a MDR/RR-TB regimen requires detailed knowledge about resistance against second-line anti-tuberculosis drugs. Performing phenotypic drug susceptibility tests (DST) for multiple anti-tuberculosis drugs is time-consuming, laborious, and for some drugs impossible.
Next generation sequencing (NGS) techniques are on the verge to replace phenotypic methods by predicting drug susceptibility and resistance based on mutation catalogs. NGS can be performed on direct patient specimen and results are available within days. Advances in machine learning (ML) techniques allow unbiased predictions from genotype to phenotype within minutes. We combined state- of-the-art expertise in ML and genome-based resistance diagnostics to develop an engine that predicts drug resistance profiles from whole genome sequencing (WGS) data.