digIS: towards detecting distant and putative novel insertion sequence elements in prokaryotic genomes

BMC Bioinformatics

Table 3 Detailed analysis of false positives of digIS, ISEScan, OASIS, and ISsaga on manually curated

Tool	Common metrics		Detailed classification of mFPs				Modified metrics
Tool	TP	FP	mFP	eIS	pNov	nIS	teIS	pNovDR (%)	nISDR (%)	pNov/nIS
Dataset ISbrowser (N = 1192)
OASIS	791	77	75	59	6	10	850	0.69	1.15	0.60
ISEScan	925	94	94	69	3	22	993	0.29	2.16	0.14
ISEScan-fragments	1077	248	239	103	18	118	1179	1.37	8.97	0.15
ISsaga	1135	363	323	148	31	144	1282	2.13	9.88	0.22
digIS	979	194	194	130	22	42	1108	1.88	3.58	0.52
Dataset E. coli (N = 49)
OASIS	26	4	4	4	0	0	30	0.00	0.00	0.00
ISEScan	43	8	7	3	0	4	46	0.00	8.00	0.00
ISEScan-fragments	45	18	17	6	0	11	51	0.00	17.74	0.00
ISsaga	48	29	28	10	0	18	58	0.00	23.68	0.00
digIS	43	11	11	6	0	5	49	0.00	9.26	0.00

N represents the number of outputs found by the tool; mFP represents the number of False Positives after merging fragments or ORFs referencing the same IS element; eIS, pNov, and nIS represent the number of mFPs classified into categories IS element with a high level of evidence, Distant or putative novel IS element, and Improbable or not an IS element, respectively; pNovDR is putative Novel Discovery Rate; nISDR is Improbable or not an IS element Discovery Rate, and pNov/nIS shows the ratio between the number of putative novel IS elements and improbable or not an IS elements.

ISSN: 1471-2105