ECHO

ECHO corrects base-call errors in short-read sequencing data without a reference genome, improving read accuracy for downstream analyses such as de novo assembly and variant detection.

Key Features:

• Reference-free operation: Performs error correction without using a reference genome.
• Parameter optimization: Automatically sets parameters based on an assumed probabilistic model tailored to each sequencing run.
• Quality scoring: Assigns a quality score to each corrected base.
• Heterozygosity modeling: Explicitly models heterozygous sites in diploid genomes to identify bases originating from those sites.
• Coverage-aware performance: Improves error-correction accuracy across varying sequence coverage depths and particularly at the ends of reads.
• De novo assembly facilitation: Serves as a preprocessing step to improve input read quality for de novo genome assembly, especially under limited coverage.

Scientific Applications:

• Genome assembly: Provides cleaner short-read data to improve accuracy of de novo genome assembly.
• Variant calling: Enhances detection of variants by modeling heterozygous sites and correcting sequencing errors.
• Population genomics: Enables reference-free analyses across diverse populations or species and handles nonuniform coverage.

Methodology:

ECHO applies a reference-free error-correction algorithm that explicitly models heterozygous sites in diploid genomes, uses an assumed probabilistic model to automatically set parameters per sequencing run, and assigns quality scores to corrected bases.