HyperGen

HyperGen leverages hyperdimensional computing to encode genomes into quasi-orthogonal vectors (Hypervector, HV) in high-dimensional space, enabling efficient and accurate Average Nucleotide Identity (ANI) estimation for large-scale genome collections through compact genome sketching.

Key Features:

• Hypervector Genome Encoding: HyperGen converts genomes into quasi-orthogonal Hypervectors in high-dimensional space rather than discrete k-mer hashes, preserving more information for accurate ANI estimation.
• Compact Sketch Representation: HV sketch representation reduces required sketch sizes while maintaining high ANI estimation accuracy and producing memory-efficient sketch files.
• Vector Multiplication-based ANI Estimation: ANI estimation is performed using vector multiplication, which naturally benefits from highly optimized general matrix multiply (GEMM) routines.
• Scalable Performance: HyperGen achieves comparable or superior ANI estimation error and linearity compared to other sketch-based counterparts, with linear scalability for massive genome collections.

Scientific Applications:

• Large-Scale Genome Similarity Estimation: HyperGen enables efficient ANI estimation for massive genome collections where exact computation incurs prohibitive runtime overhead.
• Genome Database Search: HyperGen supports fast database search operations for sketch-based genome similarity queries across datasets at various scales.

Methodology:

HyperGen encodes genomes using hyperdimensional computing to generate quasi-orthogonal Hypervectors in high-dimensional space, with ANI estimation performed via vector multiplication leveraging GEMM routines.