Table II. Summary of supporting evidence for existing gene models.

For each annotated gene model, the fraction of transcript length covered by at least one EPT at each base position was calculated using bedtools. R (38) was used to transform the resulting coverage counts into a cumulative frequency table. Fractions listed represent the lower edge of bins, exclusive. The total number of predicted gene models is likely to be artificially high due to splitting of genes over multiple small short read contigs.

Fraction of transcript covered	Cumulative abs. frequency	Cumulative percent
0.9	18	0.03
0.8	160	0.25
0.7	533	0.83
0.6	1261	1.97
0.5	2205	3.44
0.4	3475	5.42
0.3	5031	7.84
0.2	7164	11.17
0.1	10436	16.27
>0.0	15541	24.23
All	64152	100.00