Transmembrane helix prediction using amino acid property features and latent semantic analysis

BMC Bioinformatics

Table 4 Evaluation of TMHMM, SOSUI, DAS TMfilter and TMpro NN prediction performance on PDBTM non-redundant set and MPtopo high resolution set.

			F	Qhtm	Qhtm
	Method	Q _ok	Score	%obs	%prd	Q2	Confusion with soluble
PDBTM (191 proteins, 789 TM segments)
1	TMHMM	68	90	89	90	84	13
2	SOSUI	60	87	86	87		17
3	DAS TMfilter	62	90	90	91	85	10
4	TMpro NN	57	93	93	93	81	2
Without Singular Value Decomposition (SVD)
5	TMpro no SVD	57	91	93	90	81
MPtopo (101 proteins, 443 TM segments)
6	TMHMM	66	91	89	94	84	5
7	SOSUI	68	89	91	87	82	7
8	DAS TMfilter	66	88	87	90	78	5
9	TMpro NN	60	93	92	95	79	1

For description of columns, see caption of Table 3. Q^htm,%obs and Q^htm,%pred have been computed per-protein and averaged over all the proteins. Last column shows the number of membrane proteins that have been mistaken as soluble proteins.

ISSN: 1471-2105