BMC Bioinformatics

official impact factor 3.03
Search for
Advanced search
BMC Bioinformatics
Tools
Share this article
Open Access Highly Access Research article

Using quality scores and longer reads improves accuracy of Solexa read mapping

Andrew D Smith†, Zhenyu Xuan† and Michael Q Zhang*

  • * Corresponding author: Michael Q Zhang mzhang@cshl.edu

  • † Equal contributors

Author Affiliations

Cold Spring Harbor Laboratory, Cold Spring Harbor, NY 11274, USA

For all author emails, please log on.

BMC Bioinformatics 2008, 9:128 doi:10.1186/1471-2105-9-128

Published: 28 February 2008

Abstract

Background

Second-generation sequencing has the potential to revolutionize genomics and impact all areas of biomedical science. New technologies will make re-sequencing widely available for such applications as identifying genome variations or interrogating the oligonucleotide content of a large sample (e.g. ChIP-sequencing). The increase in speed, sensitivity and availability of sequencing technology brings demand for advances in computational technology to perform associated analysis tasks. The Solexa/Illumina 1G sequencer can produce tens of millions of reads, ranging in length from ~25–50 nt, in a single experiment. Accurately mapping the reads back to a reference genome is a critical task in almost all applications. Two sources of information that are often ignored when mapping reads from the Solexa technology are the 3' ends of longer reads, which contain a much higher frequency of sequencing errors, and the base-call quality scores.

Results

To investigate whether these sources of information can be used to improve accuracy when mapping reads, we developed the RMAP tool, which can map reads having a wide range of lengths and allows base-call quality scores to determine which positions in each read are more important when mapping. We applied RMAP to analyze data re-sequenced from two human BAC regions for varying read lengths, and varying criteria for use of quality scores. RMAP is freely available for downloading at http://rulai.cshl.edu/rmap/ webcite.

Conclusion

Our results indicate that significant gains in Solexa read mapping performance can be achieved by considering the information in 3' ends of longer reads, and appropriately using the base-call quality scores. The RMAP tool we have developed will enable researchers to effectively exploit this information in targeted re-sequencing projects.