Table 2.

Accuracy of REANNOTATE's inferences relative to manual annotation

	Defragmentation		Nesting	Time
	Sensitivity	Specificity
maize (AF123535.1)	97.8%	100.0%	100.0%	100.0%
wheat (AF459639.1)	96.0%	100.0%	93.3%	90.9%

Accuracy of predictions in the Defragmentation layer of re-annotation is given by their sensitivity and specificity according to the formulas $\frac{TP}{TP+FN}$ and $\frac{TN}{TN+FP}$ respectively, where TP is the count of true positives, TN true negatives, FP false positives, and FN the count of false negatives (see Implementation) relative to the manual annotations. Here a 'prediction' refers to a sequence similarity hit reported by REPEATMASKER that has been defragmented into a TE model by REANNOTATE. For the Nesting Structure and Time layers accuracy is given as the proportion of predictions in agreement with the original manual annotation [45,51].