Fig. 2

The core algorithm of MiteFinderII in the MITE detection. a A whole genome was cut into many pieces of sequence fragments and the build of k-mer index for each fragment. b Search for sequences seeds by using k-mer index. Each seed requires at least 10bp inverted repeats (TIR structure); c Extend these seeds to a complete MITE structure by merging smaller ones, including TSD structure; d Calculate the likelihood score for each MITE candidate using log-ratio model, filter away these candidates with low score; e filter these candidates with similar sequences in the flanking regions. f Select out the represent sequences of MITE families