SCALCE

SCALCE applies a Locally Consistent Parsing-based boosting scheme to reorder high-throughput sequencing reads and improve compression efficiency and processing speed for FASTQ HTS data while preserving random-access capabilities.

Key Features:

• Locally Consistent Parsing boosting scheme: SCALCE reorders reads using a boosting scheme based on Locally Consistent Parsing tailored for genomic sequence data with limited alphabet size and high read similarity.
• Enhanced compression rate for reads: SCALCE achieves up to 4.19× better compression of sequencing reads compared to gzip.
• Improved running time via reordering: Read reordering can reduce gzip running time by a factor of 15.06 on single-core PCs with 6 GB memory, and SCALCE combined with gzip reduces running time by 2.09× compared to gzip alone.
• Comprehensive FASTQ component compression: SCALCE compresses quality scores using order-3 Arithmetic Coding (AC) and read names with gzip, producing up to 3.34× improvement in compression rate and a 1.26× running-time improvement versus gzip on unordered FASTQ files.
• Comparison with BEETL: SCALCE combined with gzip achieves up to 2.01× better compression and 5.17× faster running time compared to BEETL-based approaches.
• Compressor independence and multithreading: The reordering is independent of the underlying compressor (gzip or bzip2) and supports multithreading when using pigz.

Scientific Applications:

• Reference-free compression: Compression of HTS/FASTQ data without requiring a reference genome for storage and downstream analysis.
• Data storage and transfer optimization: Reduction of storage footprint and acceleration of data transfer for large sequencing projects.
• Support for sequence similarity searches: Preservation of random-access capabilities essential for sequence similarity searches on compressed data.

Methodology:

SCALCE reorders reads using Locally Consistent Parsing (boosting scheme), compresses quality scores with order-3 Arithmetic Coding and read names with gzip, operates independently of gzip or bzip2, and supports multithreading via pigz.