Email updates

Keep up to date with the latest news and content from BMC Bioinformatics and BioMed Central.

Open Access Research article

Detecting microsatellites within genomes: significant variation among algorithms

Sébastien Leclercq12*, Eric Rivals1 and Philippe Jarne2

Author Affiliations

1 LIRMM, UMR 5506 CNRS – Université de Montpellier II, 161 rue Ada, Montpellier, France

2 CEFE, UMR 5175 CNRS – Université de Montpellier II, 1919 route de Mende, Montpellier, France

For all author emails, please log on.

BMC Bioinformatics 2007, 8:125  doi:10.1186/1471-2105-8-125

Published: 18 April 2007

Abstract

Background

Microsatellites are short, tandemly-repeated DNA sequences which are widely distributed among genomes. Their structure, role and evolution can be analyzed based on exhaustive extraction from sequenced genomes. Several dedicated algorithms have been developed for this purpose. Here, we compared the detection efficiency of five of them (TRF, Mreps, Sputnik, STAR, and RepeatMasker).

Results

Our analysis was first conducted on the human X chromosome, and microsatellite distributions were characterized by microsatellite number, length, and divergence from a pure motif. The algorithms work with user-defined parameters, and we demonstrate that the parameter values chosen can strongly influence microsatellite distributions. The five algorithms were then compared by fixing parameters settings, and the analysis was extended to three other genomes (Saccharomyces cerevisiae, Neurospora crassa and Drosophila melanogaster) spanning a wide range of size and structure. Significant differences for all characteristics of microsatellites were observed among algorithms, but not among genomes, for both perfect and imperfect microsatellites. Striking differences were detected for short microsatellites (below 20 bp), regardless of motif.

Conclusion

Since the algorithm used strongly influences empirical distributions, studies analyzing microsatellite evolution based on a comparison between empirical and theoretical size distributions should therefore be considered with caution. We also discuss why a typological definition of microsatellites limits our capacity to capture their genomic distributions.