BLESS

BLESS corrects sequencing errors in next-generation sequencing (NGS) reads using a Bloom filter to reduce memory usage and improve downstream analyses such as alignment and de novo assembly.

Key Features:

• Single minimum-sized Bloom filter: Uses a single minimum-sized Bloom filter as the core data structure for error detection and correction.
• Low memory footprint: Achieves an average 40-fold reduction in memory usage compared to existing methods.
• False-positive tolerance: Tolerates higher Bloom filter false-positive rates without compromising correction accuracy.
• End-of-read correction: Adopts assembler-inspired principles to correct errors at the ends of reads.
• Empirical alignment improvement: Converted 69% of initially unaligned reads into correctly aligned reads after correction.
• Empirical assembly improvement: Produced de novo assemblies that were 50% longer and exhibited a 66% reduction in assembly errors.
• Validated on diverse data: Evaluated using both real and simulated sequencing data.
• Targeted data type: Designed for high-throughput sequencing (NGS) reads and large genomic datasets.

Scientific Applications:

• Read alignment improvement: Increasing correct alignment rates of NGS reads by correcting sequencing errors prior to mapping.
• De novo assembly enhancement: Improving contiguity and reducing assembly errors in de novo genome assemblies after error correction.
• Large-scale dataset processing: Enabling error correction of large genomes on systems with limited memory resources.

Methodology:

BLESS employs a single minimum-sized Bloom filter and an assembler-inspired approach to identify and correct sequencing errors, including at read ends, while tolerating elevated Bloom filter false-positive rates.