Integrating standardized whole genome sequence analysis with a global Mycobacterium tuberculosis antibiotic resistance knowledgebase

Matthew Ezewudo,Debra Hanna,Christopher Gilpin,Iñaki Comas,Leen Rigouts,Timothy C Rodwell,Thomas A Kohl ,Álvaro Chiner‐Oms ,Daniela María Cirillo ,Sébastien Gagneux ,Derrick W Crook ,Ruth Mcnerney,Robin M Warren,Amanda Borens,Christoph Lange,Camilla Rodrigues,Paolo Miotto,Matteo Zignol,Leonid Chindelevitch,David M Engelthaler,Marco Schito,Rumina Hasan,David Alland,Richard Liwski,Stefan Niemann,Angela M Starks,Keertan Dheda,Sven Hoffner,James E Posey

doi:10.1038/s41598-018-33731-1

Abstract

Drug-resistant tuberculosis poses a persistent public health threat. The ReSeqTB platform is a collaborative, curated knowledgebase, designed to standardize and aggregate global Mycobacterium tuberculosis complex (MTBC) variant data from whole genome sequencing (WGS) with phenotypic drug susceptibility testing (DST) and clinical data. We developed a unified analysis variant pipeline (UVP) (https://github.com/CPTR-ReSeqTB/UVP) to identify variants and assign lineage from MTBC sequence data. Stringent thresholds and quality control measures were incorporated in this open source tool. The pipeline was validated using a well-characterized dataset of 90 diverse MTBC isolates with conventional DST and DNA Sanger sequencing data. The UVP exhibited 98.9% agreement with the variants identified using Sanger sequencing and was 100% concordant with conventional methods of assigning lineage. We analyzed 4636 publicly available MTBC isolates in the ReSeqTB platform representing all seven major MTBC lineages. The variants detected have an above 94% accuracy of predicting drug based on the accompanying DST results in the platform. The aggregation of variants over time in the platform will establish confidence-graded mutations statistically associated with phenotypic drug resistance. These tools serve as critical reference standards for future molecular diagnostic assay developers, researchers, public health agencies and clinicians working towards the control of drug-resistant tuberculosis.

Highlights

Drug-resistant tuberculosis poses a persistent public health threat
The unified analysis variant pipeline (UVP) consists of four broad sequential steps: 1) QC of the input sequence file, 2) mapping to the reference genome, 3) variant calling and annotation, and 4) Mycobacterium tuberculosis complex (MTBC) lineage classification
Each isolate whole genome sequencing (WGS) reads is assessed for MTBC species specificity using Kraken[21], and 90% of the reads had to map to MTBC for acceptance

Summary

Introduction

The ReSeqTB platform is a collaborative, curated knowledgebase, designed to standardize and aggregate global Mycobacterium tuberculosis complex (MTBC) variant data from whole genome sequencing (WGS) with phenotypic drug susceptibility testing (DST) and clinical data. The aggregation of variants over time in the platform will establish confidence-graded mutations statistically associated with phenotypic drug resistance These tools serve as critical reference standards for future molecular diagnostic assay developers, researchers, public health agencies and clinicians working towards the control of drug-resistant tuberculosis. Effective control of TB and a reduction in transmission of drug-resistant strains globally is dependent on the rapid detection and timely availability of drug susceptibility information for MTBC isolated from patients These data are critical for effective antibiotic drug regimen selection and treatment monitoring which helps prevent transmission of drug resistant disease and achieve treatment success[4]. The combination of slow MTBC replication and the challenges of growth-based phenotypic testing, results in delays of up to several weeks before clinicians receive laboratory results and can introduce a new drug combination for patients with increasingly complex resistance profiles[5,6]

Methods

Results

Conclusion