Table 1.

Accuracy of four methods of gene prediction using the O. sativa genome

		Seqpinga	MAKER2b	GlimmerHMMc	AUGUSTUSd, e
BUSCO		95.92%	92.26%	91.53%	88.70%
CDS structure	Sn	0.6175	0.5193	0.4394	0.4717
	Sp	0.5120	0.4922	0.2774	0.3008
	Acc	0.5648	0.5058	0.3584	0.3863
Exon structure	Sn	0.4820	0.4028	0.3089	-
	Sp	0.4116	0.3880	0.2129	-
	Acc	0.4468	0.3954	0.2609	-
Nucleotide Level	Sn	0.6906	0.5950	0.6597	0.6581
	Sp	0.6484	0.6680	0.4381	0.3698
	Acc	0.6695	0.6315	0.5489	0.5140

^aSeqping: Trained using rice transcriptome; ^bMAKER2: SNAP’s rice HMM, AUGUSTUS’s maize model and rice transcriptome; ^cGlimmerHMM: Trained using rice transcriptome; ^dAUGUSTUS: maize model

^eUsing the available maize models, AUGUSTUS does not predict exon structure