Kpath

Kpath compresses short-read next-generation sequencing (NGS) data by encoding kmer paths to reduce storage, transmission, and archiving requirements for sequencing datasets.

Key Features:

• Hybrid Compression Approach: Combines reference-based and reference-free (de novo) compression via path encoding to leverage the efficiency of reference methods and the flexibility of de novo encoding.
• Path Encoding Methodology: Represents sequencing data as paths and encodes them using an analogy to context-dependent arithmetic coding to compactly represent kmers.
• Compact Kmer Storage: Implements a system for compactly storing sets of kmers to support path encoding and reduce representation size.
• Efficiency and Performance: Achieves a 3–11% reduction relative to raw sequence data stored in FASTA files and on average yields more than 34% smaller compressed datasets compared to competing methods.
• Robustness Against Reference Mismatch: Maintains effective compression even when the reference genome poorly matches the reads being encoded.

Scientific Applications:

• RNA-seq Data Compression: Validated for encoding RNA sequencing (RNA-seq) reads for transcriptomics datasets.
• Large-Scale Genomic Studies: Reduces storage and transmission requirements of NGS data to facilitate large-scale genomic studies.

Methodology:

Encodes sequencing data as paths in de Bruijn graphs, applies context-dependent arithmetic coding to compactly encode kmers, and uses a compact storage system for kmer sets while integrating reference-based and reference-free encoding.