Table 3.

High-confidence predictions for known trans-splicing regions

Distance from known site (nucleotides)	Number of regions with sites predicted (n = 17 regions)
Exact matches	12.6 (7-16)
10	1.38 (1-2)
25	1.5 (1-3)
50	1.38 (1-2)
Missing predictions	4.5 (1-8)

Overall performance of the method on a set of known trans-splicing regions (tenfold cross-validation of 214 EST mapped trans-splicing sites). Each test dataset had on average 21.5 known trans-splicing regions, of which on average 17 had predictions. Missing predictions indicate those sequences for which no high-confidence prediction was available or where the nearest prediction was more than 50 nucleotides away. The mean of all ten datasets is reported with the range of values in parentheses.