Article Preview
TopIntroduction
A tandem repeat in DNA is two or more contiguous, approximate copies of a pattern of nucleotides. Tandem repeats, also called satellite DNA, are widespread in the human genome, and to date, have been linked to over 30 hereditary disorders in humans, including fragile X syndrome, myotonic dystrophy, Huntington's disease, various spinocerebellar ataxias, Friedreich's ataxia, and others(Gatchel & Zoghbi, 2005). These diseases are sometimes called trinucleotide repeat expansion diseases since they are caused by long and highly polymorphic tandem repeats of period size 3 (Mirkin, 2008). DNA forensics is another major area in which tandem repeats are essential (Jeffreys, 1993). In recent findings, microsatellites, i.e. tandem repeats with period sizes 1-6, have been shown to distinguish species (Galindo et al., 2009) and to play an important role in cancer biology (Galindo et al., 2011).
Several software tools are available for finding tandem repeats in a sequence, some of which have been used to construct databases of tandem repeats. TRF (Benson, 1999) is the basis of TRDB (Gelfand, Rodriquez, & Benson, 2007). TRed (Sokol, Benson, & Tojeira, 2007) is the software used in the TRedD database (Sokol & Atagun, 2010). Other software tools include mreps (Kolpakov & Kucherov, 2003) and ATRHunter (Wexler, Yakhini, Kashi, & Geiger, 2004). A newly developed tandem repeat meta-search engine, TReads (Pellegrini, Renda, & Vecchio, 2012), allows a user to run several of the above software tools on a given sequence with similar parameters.
The multiplicity of tandem repeat finding software stems from the fact that tandem repeats in biological sequences are approximate repeats and that there are many different ways of characterizing “fuzziness” in a repeat. Therefore, each of these software tools is based upon certain assumptions. However, most of the tools are somewhat flexible in that it is possible to modify parameters and affect the set of reported repeats.
The approaches taken by these tools can be divided into two general categories. The first is a consensus-type approach, based upon the hypothesis that there is some string called a consensus, which is similar to all copies in the repeat, but is not necessarily an exact match to any actual copy. This approach yields a multiple alignment of the copies in the tandem repeat with the number of columns equal to the length of the consensus. Benson et al. in TRF (Gelfand, Rodriquez, & Benson, 2007) follow the consensus approach.
TRedD (Sokol, Benson, & Tojeira, 2007; Sokol & Atagun, 2010) uses another approach, based upon evolutive tandem repeats (Groult, Leonard, & Mouchard, 2004). The assumption is that each copy is derived from a neighboring copy, possibly with mutations. Thus, each copy in the repeat is similar to its predecessor and successor copy, but there is not necessarily a consensus over all copies.
We point out that both mreps (Kolpakov & Kucherov, 2003) and ATRHunter (Wexler, Yakhini, Kashi, & Geiger, 2004) allow either the consensus or evolutive approach, and this is accomplished by using particular parameter settings.
While each tool offers its unique insight into the repetitive sequences in a genome, very little effort has been put into the annotation and usability of the findings. The nature of tandem repeats, including their abundance, the presence of mutations, and rotational equivalence (TTATTATTA could be reported as TTA, TAT or ATT) makes this a difficult task. For example, in Chromosome 1 of Homo Sapiens, TRF locates 72,530 repeats, and TRedD locates 91,814 repeats. As shown in Galindo et al. (2009) and Galindo et al. (2011), it is critical to have the ability to study the global content of tandem repeats across an entire genome. To study the global content experimentally requires customized arrays that can only target a finite number of repeats. Our goal is to automate the classification of tandem repeats in a manner that facilitates the study of tandem repeats across an entire genome.