Seqping: gene prediction pipeline for plant genomes using self-training gene models and transcriptomic data

BMC Bioinformatics

Table 2 Accuracy of four methods of gene prediction using the A. thaliana genome

		Seqping^a	MAKER2^b	GlimmerHMM^{c, e}	AUGUSTUS^d
BUSCO		96.44%	94.14%	98.12%	98.64%
CDS Structure	Sn	0.2749	0.0738	0.2804	0.3075
	Sp	0.8867	0.8877	0.7515	0.7527
	Acc	0.5808	0.4808	0.5160	0.5301
Exon structure	Sn	0.4596	0.1207	-	0.4155
	Sp	0.7313	0.7457	-	0.5373
	Acc	0.5955	0.4332	-	0.4764
Nucleotide Level	Sn	0.7929	0.1932	0.9350	0.9634
	Sp	0.9748	0.9750	0.8150	0.7974
	Acc	0.8839	0.5841	0.8750	0.8804

^aSeqping: Trained using A. thaliana transcriptome; ^bMAKER2: SNAP’s A. thaliana HMM, AUGUSTUS’s A. thaliana model and A. thaliana transcriptome; ^cGlimmerHMM: A. thaliana model; ^dAUGUSTUS: A. thaliana model
^eUsing the available A. thaliana model, GlimmerHMM does not predict exon structure

ISSN: 1471-2105