RNAspa: a shortest path approach for comparative prediction of the secondary structure of ncRNA molecules

BMC Bioinformatics

Table 1 Comparison of RNAspa's accuracy versus that of other prediction programs

Family	Num. of seed seq. in Rfam	Num. of datasets of ten	Avg. sequence identity	Avg. length and STD	MCC
					RNA- subopt's first choice	RNA- spa	Stem- Loc	pmMulti	Fold-AlignM	RNAcast
tRNA	1114	5	43%	71 ± 4	0.64	0.86	0.61	0.73	0.95	0.45 (3)
Tymo	28	2	68%	79 ± 3	0.76	0.93	0.98	0.72	0.94	0.92 (1)
SAM	71	5	62%	111 ± 15	0.59	0.66	0.38	0.64	0.53	0.62 (1)
Lysine	60	5	50%	179 ± 7	0.63	0.68	0.28	0.64	0.52 (1)	0.71 (1)
Purine	37	3	56%	96 ± 1	0.68	0.86	0.51	0.74	0.78	0.72 (1)
Cobalamin	171	5	46%	200 ± 27	0.38	0.40	0.12	no data	no data	0.41
FMN	48	4	60%	135 ± 19	0.32	0.44	0.25	0.45	0.64	0.47 (1)
glmS	14	1	51%	159 ± 40	0.58	0.61	0.38	no data	no data	no data

Each family was represented by several non-overlapping datasets (see third column). The MCC score is averaged among the datasets of each family. The number of datasets in which the program failed to produce an output appears in brackets. "no data" denotes that the program failed on all datasets. In some cases, StemLoc returns a consensus structure for only part of a dataset. Therefore, the MCC score reported here only reflects those sequences for which a structure was predicted. See [Additional File 5] for raw data.

ISSN: 1471-2105