A hybrid and scalable error correction algorithm for indel and substitution errors of long reads

Area of Science:

• Genomics
• Bioinformatics
• Computational Biology

Background:

• Long-read sequencing offers advantages for genome assembly over short-read sequencing.
• However, long reads have higher error rates and costs, posing computational challenges.
• Existing methods struggle with the scale and accuracy required for comprehensive long-read analysis.

Purpose of the Study:

• To introduce ParLECH, a novel hybrid error correction tool for long sequencing reads.
• To leverage high-throughput short-read data for accurate long-read error correction.
• To develop a scalable and efficient solution for processing large-scale genomic datasets.

Main Methods:

• ParLECH employs a hybrid methodology utilizing k-mer coverage information from short reads.
• It constructs a de Bruijn graph from short reads to correct indel errors in long reads.
• Substitution errors are rectified using a majority voting approach based on short-read k-mer coverage.

Main Results:

• ParLECH demonstrates superior performance compared to state-of-the-art hybrid error correction methods.
• Accurate and scalable correction of large-scale PacBio datasets, including human genome data.
• Achieved over 92% base alignment accuracy for an E. coli dataset against its reference genome.

Conclusions:

• ParLECH offers a novel hybrid error correction approach for both indel and substitution errors.
• The tool is highly scalable, capable of processing terabytes of sequencing data.
• ParLECH enhances the utility of long-read sequencing for complex genomic analyses.