Accurate self-correction of errors in long reads using de Bruijn graphs.

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Long read sequencing technologies (PacBio SMRT, Oxford Nanopore) offer long reads (up to 50,000 bp) but have high error rates (≥15%).
• Accurate long reads are crucial for applications like de novo genome assembly.
• Existing error correction methods rely on read-to-read alignment or hybrid approaches using short reads.

Purpose of the Study:

• To develop and evaluate an error correction method that exclusively uses long reads.
• To improve the accuracy and efficiency of long read sequencing data processing.

Main Methods:

• An alignment-free, iterative correction method using de Bruijn graphs with increasing k-mer lengths.
• A polishing step utilizing long-distance dependencies identified through multiple alignments.

Main Results:

• The proposed method (LoRMA) demonstrates superior accuracy compared to other long-read-only approaches for high-coverage datasets.
• At ≥75× coverage, LoRMA achieves at least 20% higher throughput.
• The method is freely available as LoRMA.

Conclusions:

• LoRMA provides a highly accurate and efficient solution for long read error correction without requiring short reads.
• This advancement facilitates more reliable de novo genome assembly and other downstream analyses using long sequencing data.