Mantis

Mantis indexes large collections of raw-read sequencing experiments using counting quotient filters to provide exact, space-efficient sequence search and colored de Bruijn graph representation.

Key Features:

• Space Efficiency: Uses a space-efficient data structure that supports indexing thousands of raw-read experiments for large-scale sequence searches.
• Exact Indexing: Employs counting quotient filters instead of traditional Bloom filters to produce exact indexes without false positives or negatives.
• Colored de Bruijn Graph Representation: Represents indexed data as a colored de Bruijn graph to enable more complex comparative analyses.
• Scalability and Updatability (Dynamic Mantis): Dynamic Mantis incorporates the Bentley–Saxe transformation to support efficient updates and scalable index growth.
• Performance Benchmarks: In comparative studies, Dynamic Mantis scaled to approximately 40,000 Sequence Read Archive (SRA) samples versus ~5,000 for VariMerge and ~80 for Bifrost.
• Efficiency in Construction and Querying: Dynamic Mantis reported faster index-construction time, lower memory usage, queries over 24× faster than Bifrost, and indexes ~2.5× smaller than Bifrost and about half the size of VariMerge.

Scientific Applications:

• Large-scale sequence search: Enables exact querying across thousands to tens of thousands of raw-read sequencing experiments.
• Comparative genomics and sequence analysis: Supports analyses based on colored de Bruijn graphs for detailed sequence comparisons and related genomic investigations.
• Variant calling and advanced analyses: Provides an exact index foundation useful for downstream tasks such as variant calling and other advanced bioinformatics workflows.

Methodology:

Mantis builds exact indexes using counting quotient filters, and Dynamic Mantis applies the Bentley–Saxe transformation to enable efficient index updates.