Automatic annotation of eukaryotic genes, pseudogenes and promoters

Table 3 Performance data for annotating 44 ENCODE sequences by either mRNA and protein supported or ab initio predictions

	mRNA + protein supported, Sn/Sp (%)	Ab initio, Sn/Sp (%)
Nucleotide level	91.14/89.54	88.44/74.46
CDS EXACT	77.19/86.48	67.54/64.22
CDS OVERLAP	90.60/91.4	85.00/71.71
SoftBerry transcripts identical to HAVANA transcripts*	56.37% (of 612)	14.75% (of 590)

*Transcripts identical means that their CDS parts (including protein coding exons and coding parts of 5' and 3' exons) are identical; the percentage of SoftBerry transcripts is relative to the number of Softberry transcripts predicted in the corresponding category: mRNA supported, protein supported or ab initio.

ISSN: 1474-760X