LiCoRNA: aLignment of Complex RNAs

Benchmark

The dataset we used for testing and analysis is from the pseudoknots families in Rfam 11.0 families. There are 68 families and we selected 21 families as the benchmark according to the following criterion.

• Maximum sequence length is less than 200.
• Number of seed sequence in one family is more than 10.
• For families with number of seed sequence more than 20, we use the tool rnazSelectSeqs.pl to select 15-maximall-divergent sequences.

 

Family	Description	Download
RF00041	Enteroviral 3' UTR element	RF00041.txt
RF00094	Hepatitis delta virus ribozyme	RF00094.txt
RF00140	Alpha operon ribosome binding site	RF00140.txt
RF00165	Coronavirus 3' UTR pseudoknot	RF00165.txt
RF00176	Tombusvirus 3' UTR region IV	RF00176.txt
RF00233	Tymovirus/Pomovirus/Furovirus tRNA-like 3' UTR element	RF00233.txt
RF00381	Antizyme RNA frameshifting stimulation element	RF00381.txt
RF00507	Coronavirus frameshifting stimulation element	RF00507.txt
RF00622	Mammalian CPEB3 ribozyme	RF00622.txt
RF01072	Pseudoknot of upstream pseudoknot domain of the 3'UTR	RF01072.txt
RF01093	Ma3 gene ribosomal frameshift signal	RF01093.txt
RF01099	Pseudoknot of influenza A virus gene	RF01099.txt
RF01689	AdoCbl variant RNA	RF01689.txt
RF01704	Downstream peptide RNA	RF01704.txt
RF01725	SAM-I/IV variant riboswitch	RF01725.txt
RF01726	SAM-II long loop	RF01726.txt
RF01735	epsC RNA	RF01735.txt
RF01761	wcaG RNA	RF01761.txt
RF01786	Cyclic di-GMP-II riboswitch	RF01786.txt
RF01831	THF riboswitch	RF01831.txt
RF01840	ribosomal frameshift element	RF01840.txt

 

Result

We use three parameters to estimate the performance of LiCoRNA. Average Fractional Identity (AFI) consider only the sequence alignment. The Fractional Identity represents the alignment identity between the test and reference alignment, that is the number of identities divided by the length of the alignment. Sensitivity/Specificity consider the structure predicted by the alignment where Specificity=TP/(TP+FP) and Sensitivity = TP/(TP+FN) . TP(true positive) represents the number of correctly predicted base pair, FP(false positive) represents the number of predicted base pairs which are not in the annotated structure and FN(false negative) denotes the number of base pairs in the annotated structure that are not predicted.

 

Family	Length	Average Identity	tree-width	Sensitivity	Specificity	AFI	Download
RF00041	120 - 130	83.40%	4	0.895	0.940	0.934	RF00041.tar.gz
RF00094	87 - 93	86.11%	4	0.911	0.931	0.881	RF00094.tar.gz
RF00140	99 - 115	77.66%	4	0.903	0.941	0.826	RF00140.tar.gz
RF00165	62 - 64	68.42%	3	0.963	0.976	0.916	RF00165.tar.gz
RF00176	89 - 92	93.28%	3	0.956	0.960	0.973	RF00176.tar.gz
RF00233	78 - 86	73.72%	3	0.914	0.935	0.954	RF00233.tar.gz
RF00381	56 - 59	84.19%	3	0.950	0.991	0.966	RF00381.tar.gz
RF00507	79 - 87	73.97%	3	0.931	0.982	0.952	RF00507.tar.gz
RF00622	74 - 78	84.50%	3	0.927	0.970	0.980	RF00622.tar.gz
RF01072	29 - 33	80.89%	3	0.973	0.999	0.974	RF01072.tar.gz
RF01093	60	76.41%	3	0.906	0.955	0.983	RF01093.tar.gz
RF01099	48	86.15%	3	0.936	1.000	1.000	RF01099.tar.gz
RF01689	125 - 134	83.26%	4	0.957	0.986	0.966	RF01689.tar.gz
RF01704	56 - 64	75.50%	3	0.981	0.999	0.933	RF01704.tar.gz
RF01725	90 - 128	84.02%	3	0.974	0.998	0.938	RF01725.tar.gz
RF01726	59 - 61	79.02%	3	0.966	0.999	0.950	RF01726.tar.gz
RF01735	112 - 125	77.23%	3	0.947	0.963	0.906	RF01735.tar.gz
RF01761	93 - 100	80.15%	3	0.970	0.980	0.933	RF01761.tar.gz
RF01786	83 - 86	64.85%	4	0.906	0.964	0.942	RF01786.tar.gz
RF01831	96 - 103	74.71%	4	0.961	0.993	0.960	RF01831.tar.gz
RF01840	54	88.14%	3	0.945	0.991	0.995	RF01840.tar.gz