Minos

Minos adjudicates variant calls by integrating outputs from multiple variant callers using genome-graph read mapping to resolve SNP and indel discrepancies and improve genotyping accuracy in bacterial cohorts.

Key Features:

• Statistical adjudication: Minos applies a statistical approach to adjudicate variants at loci where different variant calling tools disagree.
• Genome graph mapping: It maps reads onto a genome graph constructed from variant calls to enable integration of variant outputs.
• Joint genotyping: It frames joint genotyping of SNPs and indels within bacterial cohorts as an adjudication problem to provide unified genotypes.
• Integration without loss of precision: Minos integrates outputs from multiple variant callers without loss of precision.
• Non‑SNP and distant reference handling: Genome-graph mapping accommodates non-SNP variations and can handle potentially distant reference genomes more effectively than single-reference methods.

Scientific Applications:

• Cross-species benchmarking: It has been benchmarked across bacterial species including Mycobacterium tuberculosis, Staphylococcus aureus, and Klebsiella pneumoniae.
• Outbreak investigation: Applied in an outbreak investigation involving 385 Mycobacterium tuberculosis samples.
• Large-cohort rifampicin analysis: Used in a cohort analysis of approximately 15,000 samples with known rifampicin phenotypes.
• RRDR variant mapping: Used to build comprehensive maps of non-synonymous variants within the rifampicin resistance determining region (RRDR) of the rpoB gene and to correlate specific RRDR SNPs with rifampicin resistance levels.
• Replication cohort: Findings were replicated in a second cohort of approximately 13,000 samples.

Methodology:

Minos integrates multiple variant calling outputs by constructing a genome graph from variant calls, mapping reads to that graph, and applying a statistical adjudication procedure to resolve discrepancies and perform joint genotyping of SNPs and indels.