RabbitSketch

RabbitSketch implements a high-performance sketching library for genome analysis that exploits modern multi-core CPUs.

Key Features:

• Sketching Algorithms: Implements MinHash, OrderMinHash, HyperLogLog, and Kssd algorithms for genomic sketching.
• Multi-core CPU Optimization: Achieves speedups ranging from 2.30x to 49.55x compared to existing implementations.
• Language Bindings: Provides interfaces for both Python and C++.
• Integration Capability: Integrates with tools such as RabbitTClust, as demonstrated by the Kssd algorithm integration yielding 1.54x speedup with no loss in accuracy.

Scientific Applications:

• Similarity Analysis: Enables rapid similarity analysis of large-scale genomic datasets, such as 455GB genomic data.
• Genome Clustering: Enhances genome clustering tools through algorithm integration, as shown with RabbitTClust enhancement.

Methodology:

RabbitSketch implements sketching algorithms including MinHash, OrderMinHash, HyperLogLog, and Kssd with optimization for multi-core CPU architectures.