BLight

BLight provides an exact static associative index for k-mers to enable compact storage and high-throughput querying in sequence analysis.

Key Features:

• Static and Exact Data Structure: BLight is a static associative data structure that maps unique identifiers to k-mers and guarantees no false positives in membership queries.
• Memory Efficiency: It leverages overlaps among k-mers to achieve a compact representation, reducing memory requirements to as low as 23 bits per k-mer for the human genome and 27 bits per k-mer for the axolotl genome.
• Scalability and Performance: It indexes large genomes efficiently, e.g., the human genome with 8 GB of RAM in 10 minutes and the axolotl genome with 63 GB of memory in 76 minutes.
• High Throughput Querying: BLight supports up to 1.4 million queries per second on a single CPU and up to 16.1 million queries per second using 12 cores.
• Efficient Construction: Index construction is efficient and does not require additional memory beyond what is necessary for the index itself.
• Representation of Sequencing Data: The data structure can represent metagenomic and transcriptomic sequencing data as compact k-mer indexes.

Scientific Applications:

• Genomic Research: Facilitates analysis of large genomes by providing compact, exact k-mer indexing and fast queries.
• Metagenomics: Enables representation and analysis of complex microbial community k-mer sets using compact indexes.
• Transcriptomics: Supports management and querying of transcriptome-derived k-mers for RNA sequencing analyses.

Methodology:

BLight constructs a static associative index of k-mers that leverages overlaps between k-mers to compress representation, supports exact membership queries, and is built without requiring additional memory beyond the index.