pathotypr

Pathotypr performs alignment-free lineage classification and marker-driven genotyping from whole-genome sequencing data for microbial pathogens.

Key Features:

• Pathogen-agnostic framework: Users can supply custom SNP marker panels for any microbial pathogen.
• Curated MTBC panels: Ships with 3,707 lineage-defining SNPs and over 102,000 resistance-associated mutations for Mycobacterium tuberculosis complex.
• Multiple input formats: Processes assembled genomes in FASTA format and raw reads in FASTQ format.
• Comprehensive lineage coverage: Supports all 14 currently recognised MTBC lineages (L1-L10, A1-A4).
• Resistance genotyping: Performs WHO catalogue-based resistance-associated variant detection.
• High performance: Achieves approximately 1 second runtime per sample, enabling analysis of 88,071 samples in approximately 24 hours on four threads.
• High concordance: Marker-based and alignment-free lineage calls achieved 100% concordance on complete genomes and 100% root-lineage concordance with TB-Profiler on non-ambiguous samples.
• Genotype-phenotype prediction: Resistance predictions showed 85.0% overall concordance, with 95.8% sensitivity and 95.0% specificity for rifampicin and 93.0% sensitivity and 97.9% specificity for isoniazid.

Scientific Applications:

• MTBC genomic surveillance: Enables harmonised lineage assignment and resistance screening for near real-time and cross-border tuberculosis surveillance.
• Transmission dynamics analysis: Supports reconstruction of probable introduction events, as demonstrated by identifying 135 probable introduction events into Germany, Italy and Ukraine in an MDR-enriched collection.
• Clinical resistance screening: Provides high-confidence detection of MDR and pre-XDR genotypes in clinical isolates.
• Large-scale population studies: Facilitates rapid processing of tens of thousands of samples for phylogenetic and epidemiological investigations.

Methodology:

Pathotypr implements a k-mer/Random Forest framework with marker-based lineage calling and WHO catalogue-based resistance calling. The MTBC marker backbone was derived from a phylogeny reconstructed from 26,813 genomes using 609,003 polymorphic sites.