Abstract
BackgroundHigh-throughput sequencing offers higher throughput and lower cost for sequencing a genome. However, sequencing errors, including mismatches and indels, may be produced during sequencing. Because, errors may reduce the accuracy of subsequent de novo assembly, error correction is necessary prior to assembly. However, existing correction methods still face trade-offs among correction power, accuracy, and speed.ResultsWe develop a novel overlap-based error correction algorithm using FM-index (called FMOE). FMOE first identifies overlapping reads by aligning a query read simultaneously against multiple reads compressed by FM-index. Subsequently, sequencing errors are corrected by k-mer voting from overlapping reads only. The experimental results indicate that FMOE has highest correction power with comparable accuracy and speed. Our algorithm performs better in long-read than short-read datasets when compared with others. The assembly results indicated different algorithms has its own strength and weakness, whereas FMOE is good for long or good-quality reads.ConclusionsFMOE is freely available at https://github.com/ythuang0522/FMOC.
Highlights
High-throughput sequencing offers higher throughput and lower cost for sequencing a genome
This paper presented a novel overlap-based error correction algorithm using FM-index
Given a query read, we first identify reads overlapping with the query by performing alignment against reads compressed in FM-index, construct a multiple-sequence alignment (MSA) matrix, and replace the less-frequent alleles on the query with the most-frequent one at the same locus
Summary
High-throughput sequencing offers higher throughput and lower cost for sequencing a genome. The reads generated by generation sequencing platforms (e.g., Illumina, Roche 454) may contain several types of errors including mismatches, insertions and deletions (collectively termed indels) [1]. These errors bring great challenges of subsequent genome assembly algorithms, because false read overlaps may be produced, which further leads to fragmented assembly or even misassembly. These errors will increase the size of assembly graph due to erroneous vertices and edges, which implies requirement of larger memory usage and computational time.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have