Correcting errors in short reads by multiple alignments

Abstract

Current sequencing technologies produce a large number of erroneous reads. The sequencing errors present a major challenge in utilizing the data in de novo sequencing projects as assemblers have difficulties in dealing with errors. We present Coral which corrects sequencing errors by forming multiple alignments. Unlike previous tools for error correction, Coral can utilize also bases distant from the error in the correction process because the whole read is present in the alignment. Coral is easily adjustable to reads produced by different sequencing technologies like Illumina Genome Analyzer and Roche/454 Life Sciences sequencing platforms because the sequencing error model can be defined by the user. We show that our method is able to reduce the error rate of reads more than previous methods. The source code of Coral is freely available at http://www.cs.helsinki.fi/u/lmsalmel/coral/.