Table 3.

Prediction accuracies of alignments generated by joint sampling and cluster analysis, the alignments in the minimum pseudo-free energy structural alignment as predicted by PARTS MAP algorithm (denoted by ‘PARTS MAP’) and the ML sequence alignment computed by sequence alignment hidden Markov model (denoted by ‘pHMM’)

	Biggest cluster		Ensemble		Best cluster		PARTS MAP		pHMM
	Sensitivity	PPV	Sensitivity	PPV	Sensitivity	PPV	Sensitivity	PPV	Sensitivity	PPV
tRNA	0.857	0.856	0.857	0.858	0.881	0.881	0.843	0.847	0.794	0.787
5S rRNA	0.940	0.946	0.940	0.947	0.951	0.957	0.925	0.932	0.906	0.902
RNase P	0.744	0.720	0.744	0.721	0.753	0.729	0.743	0.705	0.743	0.703
Overall	0.847	0.841	0.847	0.842	0.862	0.856	0.837	0.828	0.814	0.797

The sequence families are denoted on the rows and prediction accuracy of Biggest cluster centroid, Ensemble centroid, Best cluster centroid, PARTS MAP and pHMM (pairwise hidden Markov model) are denoted on columns.