An error correction algorithm for NGS data

Abstract

The Oxford Nanopore and Pacbio SMRT sequencing technologies has revolutionized the Next-Generation Sequencing (NGS) environment by producing long reads that exceed 60 kbp and helped to the completion of many biological projects. But, long reads are characterized by a high error rate which increases the difficulty of biological problems like the genome assembly problem. Error correction of long reads has become a challenge for bioinformaticians, which motivates the development of new approaches for error correction adapted to NGS technologies. In this paper, we present a new denovo self-error correction algorithm using only long reads. Our algorithm operates in two steps: First, we use a fast hashing method which allows to find alignments between the longest reads and other reads in a set of long reads. Next, we use the longest reads as seeds to obtain the final alignment of long reads by using a dynamic programming algorithm in a band of width w. Our error correction algorithm does not require high quality reads, in contrast to existing hybrid error correction ones.