Efficient pairwise RNA structure prediction using probabilistic alignment constraints in Dynalign

BMC Bioinformatics

Table 10 Structural prediction accuracy statistics for the methods benchmarked over 2000 random tRNA selections when scored using an exact match criterion.

		Percent sequence similarity
		20–40	40–60	60–80	80–100	0–100
Dynalign new constraint	Sens	0.818	0.862	0.923	0.715	0.861
	PPV	0.788	0.832	0.916	0.697	0.834
Dynalign M constraint	Sens	0.813	0.854	0.917	0.716	0.855
	PPV	0.781	0.822	0.907	0.698	0.825
FOLDALIGN	Sens	0.731	0.846	0.887	0.690	0.835
	PPV	0.815	0.894	0.939	0.846	0.889
StemLoc	Sens	0.581	0.894	0.940	0.872	0.860
	PPV	0.747	0.899	0.924	0.876	0.886
Consan	Sens	0,786	0.905	0.944	0.769	0.894
	PPV	0.755	0.848	0.873	0.667	0.838
Single Prediction	Sens	0.743	0.746	0.774	0.717	0.748
	PPV	0.696	0.688	0.728	0.673	0.693

Results are summarized for sequence similarity values ranging from 20% through 100% in steps of 20% and for the overall data set (0 – 100). Dynalign new constraint refers to Dynalign with probabilistic alignment constraints proposed here. Software version numbers and parameters for the algorithms are described in Section 7.6.

ISSN: 1471-2105