MegaPath

MegaPath detects pathogens from metagenomic next-generation sequencing (NGS) data by applying polishing to remove non-informative human reads and spurious alignments, global optimization and read reassignment, an enhanced maximum-exact-match prefix seeding strategy, and a SIMD-accelerated Smith-Waterman algorithm to identify novel or highly divergent bacteria and viruses.

Key Features:

• High Sensitivity in Pathogen Detection: Detects pathogens with high sensitivity even when sequence similarity to reference databases is low, enabling identification of novel or highly divergent bacteria and viruses.
• Polishing Techniques: Removes non-informative human reads and spurious alignments to improve signal-to-noise for downstream detection.
• Global Optimization and Read Reassignment: Uses a global optimization strategy to reassign reads aligned to multiple species to a unique species, increasing correct alignments to distant pathogens and reducing incorrect assignments.
• Enhanced Seeding and SIMD-accelerated Smith-Waterman: Implements an enhanced maximum-exact-match prefix seeding strategy combined with a SIMD-accelerated Smith-Waterman algorithm to accelerate alignment computations.
• Benchmark Performance: Demonstrated detection of eight times more reads from low-similarity pathogens compared to existing tools and processed a ~1 Gb dataset in approximately 20 minutes with speed comparable to profile-based methods.

Scientific Applications:

• Infectious Disease Diagnostics: Identification of bacterial and viral pathogens in clinical metagenomic NGS samples.
• Epidemiology and Public Health Response: Rapid detection of emerging pathogens to inform public health interventions.
• Pathogen Evolution and Transmission Studies: Detection of high-mutation-load and novel sequences to support analyses of pathogen evolution and transmission dynamics.
• Surveillance of Novel or Divergent Pathogens: Monitoring and detection of pathogens with substantial genetic divergence from known reference sequences.

Methodology:

Polishing to remove non-informative human reads and spurious alignments; global optimization with read reassignment to unique species; enhanced maximum-exact-match prefix seeding; SIMD-accelerated Smith-Waterman alignment.