Re-annotation algorithm. A. Graphic representation of the input REPEAT MASKER annotation of the first 100 Kb of genomic sequence [GEN BANK:AF123535.1] around the adh gene of the maize cultivar LH_82 (this refers to the same maize sequence that was manually annotated in  and used to validate REANNOTATE's predictions in Results, Table 1, and Figure 3). B. Boxes highlighted in magenta on the bottom tier represent hits to the reference element PREM2_ZM_I (the internal region of an LTR-retrotransposon in REP BASE), of which the three innermost hits, shown again in red on the top tier united by a horizontal line, were defragmented by REANNOTATE into a repetitive element model. The black arrows show the orientation of the hits on the chromosome, and the three hits shown in red are colinear with the reference PREM2_ZM_I sequence. C. Boxes highlighted in blue on the bottom tier represent hits to the reference LTR sequence PREM2_ZM_LTR (in REP BASE). Above, two (single-hit) LTR models (shown in orange) flank an IR model (in red): these three models have been assembled into a higher-order model of an element of the PREM2_ZM family. D. The chromosomal span of the defragmented PREM2_ZM element (red and orange) is within the span of another element (bottom model in black); the PREM2_ZM element is inferred to have inserted into the element shown below it. Two other elements (black boxes on top tier) are inferred to have inserted into the PREM2_ZM element. E. Pairs of intra-element LTR sequences are output, aligned with CLUSTAL W, and the number of point substitutions between them estimated.