MetWAMer: eukaryotic translation initiation site prediction

BMC Bioinformatics

Table 2 Distinguishing in-frame, upstream ATG sites from true TISs

Parametrization	Method	TP	FP	FN	Sn	Sp
	1^st-ATG	0	1,150	0	0.0000	0.0000
	TISHunter	69	1,081	0	0.0600	0.0600
	ATGpr	97	1,053	0	0.0843	0.0843
	TIS Miner	216	832	102	0.1878	0.2061
	NetStart	216	930	4	0.1878	0.1885
homogeneous	LLKR	437	671	42	0.3800	0.3944
	WLLKR	531	467	152	0.4617	0.5321
	MFCWLLKR	525	625	0	0.4565	0.4565
	PFCWLLKR	542	551	57	0.4713	0.4959
	BAYES	442	511	197	0.3843	0.4638
cluster-specific	LLKR	552	550	48	0.4800	0.5009
	WLLKR	663	380	107	0.5765	0.6357
	MFCWLLKR	683	467	0	0.5939	0.5939
	PFCWLLKR	679	419	52	0.5904	0.6184
	BAYES	567	406	177	0.4930	0.5827
random split	LLKR	434	672	44	0.3774	0.3924
	WLLKR	522	468	160	0.4539	0.5273
	MFCWLLKR	530	620	0	0.4609	0.4609
	PFCWLLKR	541	551	58	0.4704	0.4954
	BAYES	439	512	199	0.3817	0.4616

TIS identification statistics are reported exclusively for the 1,150 maximal reading frames containing in-frame ATG sites upstream from the true TIS, under three distinct parameter utilization approaches: homogeneous, a priori-known cluster-specific with k = 3, and three-fold random split. TP represents the number of instances for which the method correctly identified a TIS; FP for which a prediction was made, though incorrect; and FN for which no prediction was made, but should have been (see Figure 2). $S n = \frac{T P}{T P + F P + F N}$ , and $S p = \frac{T P}{T P + F P}$ .

ISSN: 1471-2105