ELECTOR: evaluator for long reads correction methods.

Area of Science:

• Genomics
• Bioinformatics

Background:

• Third-generation sequencing technologies produce long reads with error rates exceeding 5%, primarily insertions and deletions.
• The accuracy of long reads correction significantly impacts downstream genomic analyses.
• Existing evaluation methods for long reads correction tools are computationally expensive and do not scale well.

Purpose of the Study:

• To develop a precise, reliable, and scalable method for evaluating the performance of long reads error correction tools.
• To provide reproducible benchmarks for assessing the quality of corrected long reads, including ultra-long reads.

Main Methods:

• Introduced ELECTOR, a tool for evaluating long reads correction compatible with various correction methods.
• Implemented a novel algorithmic strategy for alignment segmentation within multiple sequence alignment to enhance scalability.
• Developed a method for reproducible correction benchmarks on ultra-long reads (>100 kb).

Main Results:

• ELECTOR provides faster performance compared to the current state-of-the-art on various datasets.
• The tool offers a comprehensive set of metrics to assess read quality improvements after correction.
• ELECTOR enables reproducible benchmarking, particularly for ultra-long sequencing reads.

Conclusions:

• ELECTOR addresses the critical need for efficient and accurate evaluation of long reads correction tools.
• The tool's scalability and comprehensive metrics make it valuable for the genomics community.
• ELECTOR facilitates reproducible and robust benchmarking of long reads correction methods for diverse sequencing applications.