A method for aligning RNA secondary structures and its application to RNA motif detection

BMC Bioinformatics

Table 1 Performance comparison of RNA secondary structure tools

Tool Name	Running Time	Space Requirement	Reference
Sankoff^a	O(N⁶)	O(N⁴)	[17]
FOLDALIGN^b	O(N⁴)	O(N⁴)	[18]
RAGA^c	O(M²N³)	O(M²N²)	[22]
rna_align^d	min{O(MN³), O(M³N)}	O(MN²)	[26]
Dynalign^e	O(M³N³)	O(M²N²)	[19]
stemloc^f	O(LM)	N/A	[32]
Rsearch^g	O(M³N)	O(M³)	[31]
RNAforester^h	O(\|F₁\|\|F₂\|deg(F₁)deg(F₂)(deg(F₁) + deg(F₂)))	O(\|F₁\|\|F₂\|deg(F₁)deg(F₂))	[27]
CARNACⁱ	O(N⁶), O(N²)	O(N⁴), O(N²)	[20]
comRNA^j	O(M^N)	N/A	[21]

^a N is the average sequence length;
^b N is the average length of a given set of RNAs;
^c M and N are the lengths of the two given sequences;
^d M and N are the two sequence lengths;
^e M is the maximum distance allowed to match two nucleotides and N is the length of the shorter sequence;
^f L and M are the two RNA sequence lengths; only valid in extreme cases;
^g M is the query length and N is the subject sequence length;
^h |F_i| is the number of nodes in forest F_i and deg(F_i) is the degree of F_i;
ⁱ N is the sequence length, theoretical time complexity of O(N⁶) could be significantly reduced to around O(N²) by pre-processing of the sequences, as noted by the authors [20].
^j M is the maximum number of stems examined and N is the number of total sequences under analysis. The comRNA's average run-time can be significantly improved by carefully chosen parameters, as noted by the authors [21].

ISSN: 1471-2105