“…Exhaustive search of TRs, instead, is unaffordable because of the exponential growth of the dictionary size as a function of the allowed motif length. The dual approach is that of ab-initio methods (Benson, 1999; Boeva et al , 2006; Girgis and Sheetlin, 2013; Kofler et al , 2007; Kolpakov et al , 2003; Krishnan and Tang, 2004; Kurtz et al , 2001; Pellegrini et al , 2010; Sokol et al , 2007; Wexler et al , 2005; Wirawan et al , 2010; Zhou et al , 2009) that do not require any pre-existing knowledge about the input sequence. Although more complex, this class of algorithms is the most studied and used in practice.…”