MAGMA

MAGMA performs comprehensive genomic analysis of clinical Mycobacterium tuberculosis (Mtb) samples to detect genomic variants associated with drug resistance, tolerance, and virulence using the XBS (compleX Bacterial Samples) framework.

Key Features:

• XBS framework: Implements the compleX Bacterial Samples (XBS) bioinformatics framework tailored for challenging Mtb samples.
• Joint calling: Uses joint variant calling to improve detection of variants across samples.
• Machine-learning-based variant filtering: Applies machine-learning-based filters to refine variant calls in complex and contaminated samples.
• Low-depth performance: Detects variants in low coverage conditions including 5–10× depth.
• Contamination robustness: Maintains performance across various contamination types and at contamination levels exceeding 50%.
• Complex genomic region detection: Improves variant detection in complex genomic regions of Mtb.
• Validation on simulated datasets: Accuracy was verified using novel simulated datasets enabling precise performance assessment.
• Comparative SNP and INDEL gains: Identified 9.0% more single nucleotide polymorphisms (SNPs) and 8.1% more single nucleotide insertions and deletions than the WHO-endorsed unified analysis variant pipeline in complex regions.
• Sputum WGS performance: Showed superior detection in sputum-like sequence data with low coverage and high contamination, identifying 13.9% more variable sites than MTBseq.
• High sensitivity on culture isolates: Reported sensitivity of 98.8% when analyzing culture isolates.
• rRNA exclusion: Excluding rRNA regions prevented generation of false positives in validated analyses.
• Benchmarked against existing pipelines: Directly compared performance to UVP, MTBseq, and the WHO-endorsed unified analysis variant pipeline.

Scientific Applications:

• Drug-resistance variant detection: Identification of genomic variants associated with drug resistance and tolerance in Mtb.
• Virulence-associated variant detection: Detection of variants potentially linked to increased virulence in Mtb.
• Direct WGS from clinical specimens: Enables whole genome sequencing directly from clinical sputum specimens to increase usable sample yield.
• Expansion of genomic datasets: Broadens the range of samples available for drug-resistance and other genomic analyses by accommodating less-than-perfect specimens.
• Genome-wide association studies (GWAS): Provides enhanced genetic resolution to support GWAS in Mtb.
• Sequence-based transmission studies: Improves variant detection for transmission and epidemiological analyses.
• Analysis of culture isolates and sputum WGS: Applicable to both cultured isolates and direct-sample (sputum) whole genome sequencing data.

Methodology:

Computational methods explicitly include the XBS framework with joint variant calling and machine-learning-based variant filtering; validation using novel simulated datasets and whole genome sequencing (WGS) clinical samples; comparative benchmarking against UVP, MTBseq, and the WHO-endorsed unified analysis variant pipeline; and exclusion of rRNA regions during variant calling.