A pragmatic pipeline for drug resistance and lineage identification in Mycobacterium tuberculosis using whole genome sequencing

Abstract

Delays in accurate diagnosis of drug resistant tuberculosis (DR-TB) can hinder treatment. Whole genome sequencing (WGS) provides more information than standard molecular and phenotypic testing, but commonly used platforms are expensive to implement, and data interpretation requires significant expertise. We aimed to optimise a TB WGS diagnostic pipeline balancing user-friendliness, cost-effectiveness and time to results, whilst ensuring accuracy. Growth conditions, DNA extraction protocols and Oxford Nanopore Technologies (ONT) library preparation kits were compared. ONT was compared with Illumina protocols. Software for basecalling and analysis were evaluated to find the most accurate resistance SNP and lineage predictor. Optimally, a spin-column CTAB DNA extraction method was combined with the RBK110.96 library preparation kit, high accuracy (HAC) basecalling and data analysis using TB-Profiler. Compared with Illumina, the pipeline was concordant for 16/17 (94%) isolates (lineage) and for 17/17 (100%) isolates (resistance SNPs). Our pipeline was 71% (12/17) concordant with phenotypic drug susceptibility test (DST) results. Time-to-diagnosis was around four weeks. This optimised TB sequencing pipeline requires less time and expertise to run and analyse than Illumina, takes less time than phenotypic DSTs and the results are comparable with Illumina. The cost per sample is comparable with other methods. These features make it an important tool for incorporating into routine DR-TB diagnostic pipelines and larger scale drug resistance surveillance in all settings.

Fig 1. Genomic DNA yield and integrity obtained from MDR-TB and DS-TB isolate cultures.

Overall, 7H11 slopes produced the greatest yield of DNA, whilst MGIT tubes produced the greatest DNA size. A) DNA yields extracted from MGIT tubes over time; there was no change in MDR-TB DNA yield, but the DS-TB isolate yield decreased. B) DNA amounts extracted from 7H11 slopes over time; there was no change in MDR-TB DNA yields, but DS-TB isolate yields decreased. C) Size of DNA extracted from MGIT tubes over time; both MDR-TB and DS-TB isolates increased over time. D) Size of DNA extracted from 7H11 slopes over time; the MDR-TB isolates peaked at week 3 and then decreased (p=0.0025), the DS-TB isolates peaked at weeks 3 and 4 (p=0.01). Significance was calculated using One-Way ANOVA and Tukey’s post ad hoc analysis.

Fig 2. Precipitation and spin-column CTAB DNA extraction method comparison.

A) DNA yield measured using Qubit dsDNA broad spectrum kit, B) Size (bp) measured using the TapeStation genomic DNA kit. Apart from spin-column CTAB from MGIT tubes, all other methods provided the minimum required yield, although precipitation based CTAB protocols provided higher yields. Spin-column CTAB from MGIT media also provided the smallest size. Significance was calculated using One-Way ANOVA and Tukey’s post ad hoc analysis.

Fig 3. DNA shearing to 10 kb and 20 kb compared to 50 kb.

A) number of reads (no significant difference) B) N50 (overall p = 0.01) and C) total number of bases (no significant difference). One-Way ANOVA and Tukey’s multiple comparison test was applied.

Fig 4. A) Mean read depth and B) genome coverage when comparing each of the three kits. Statistical significance calculated using One-Way ANOVA and Tukey’s multiple comparisons test. pCTAB = precipitation CTAB, cCTAB = spin column CTAB. Numbers in brackets denote the number of samples multiplexed per run.

Fig 5. Comparison of BAM file read depth of rpoB for

A) the PCR-based SQK-RPB004 kit, showing low coverage, B) SQK-RBK004 kit showing greater coverage and C) SQK-RBK110.96 resulted in the best coverage and depth, using Artemis.

Fig 6. Drug resistance and lineage profile concordance resulting from phenotypic DSTs, ONT and Illumina sequencing.

P = phenotypic DST results, O = Oxford Nanopore results, I = Illumina results, M = mixed (lineage 3 and 4), S = sensitive, R = resistant, NT = not tested.

Fig 7. The recommended ONT WGS TB pipeline, including culturing conditions, DNA extraction methodology, optimal DNA input library preparation kit, basecalling and data analysis.

Images taken from Windows 10 Education and ShutterStock.