SnowyOwl: accurate prediction of fungal genes by using RNA-Seq and homology information to select among ab initio models

Table 2 Effects of pipeline composition on number of gene predictions and recovery of curated models

		Components in the test pipelines (leftmost is the standard SnowyOwl pipeline)
In Consensus Training Set	Orthologs		*		*	*	*	*	*	*	*
	Contig Models	*	*	*		*	*	*	*	*	*	*	*
	GeneMark Models	*	*	*	*		*	*	*	*	*	*	*
In Pooled Augustus Models	Blat-hinted Models	*	*	*	*	*		*	*	*	*	*	*
	Unhinted Models	*	*	*	*	*	*		*	*	*	*	*
	Augustus Domains		*	*	*	*	*	*		*	*
Merged with Pooled Augustus	GeneMark Models	*	*	*	*	*	*	*	*		*	*
	Splice-hinted Models	*	*	*	*	*	*	*	*	*
		Changes in the number of predicted models relative to the standard pipeline
A. niger	Representative (21479)	0	6	61	-8	-21	-9332	12	-61	-27	13	23	-80
	Accepted (11374)	0	24	54	46	15	-1089	20	-35	-60	-5	-21	-149
T. lanuginosus	Representative (12915)	0	5	-25	-21	-45	-5367	6	-16	11	5	2	-11
	Accepted (7324)	0	28	13	17	0	-606	26	-7	0	14	-19	-64
P. chrysosporium	Representative (20201)	0	92	13	-5742	81	-6444	87	44	52	-56
	Accepted (12669)	0	111	43	-1585	43	-2058	98	48	48	-87
		Recovery of curated models (%)
A. niger	Representative	94.7	94.7	94.8	94.8	94.8	94.3	94.7	94.8	91.4	91.6	91.6	91.2
	Accepted	94.5	94.5	94.6	94.5	94.5	93.4	94.5	94.5	91.2	91.3	91.3	91.0
T. lanuginosus	Representative	93.7	93.8	93.5	93.5	93.7	91.9	93.9	94.0	92.2	92.2	92.3	92.1
	Accepted	93.6	93.7	93.4	93.4	93.6	91.6	93.8	93.9	92.1	92.1	92.2	92.0
P. chrysosporium	Representative	88.0	88.1	88.0	90.4	87.8	90.9	88.2	88.2	88.2	84.6
	Accepted	87.3	87.3	87.1	88.4	87.1	88.5	87.3	87.4	87.4	83.8

ISSN: 1471-2105