A Review on Recent Computational Methods for Predicting Noncoding RNAs

Table 3

De novo ncRNA prediction methods using RNA structure features.

Name	URL	Feature	Prediction algorithm
RNAfold [33]		Base-pair probabilities and MFE	Partition function and dynamic programming
Mfold [34]	http://unafold.rna.albany.edu/?q=mfold	MFE	Dynamic programming
Afold [35]	ftp://ftp.ncbi.nlm.nih.gov/pub/ogurtsov/Afold	Sets of conditionally optimal multibranch loop free structures	Dynamic programming
Sfold [41]	http://sfold.wadsworth.org/cgi-bin/index.pl	Internal loops, sets of conditionally optimal MLF structures	Nearest-neighbour model (NNM)
Nussinov [42]	http://www.pnas.org/content/77/11/6309	Individual base pairs and loop structure with the lowest free energy	Dynamic programming
Partition function method [43]	http://www.ncbi.nlm.nih.gov/pubmed/1695107	Full equilibrium partition for secondary structure and the probabilities of various substructures	Dynamic programming
Zhang [44]	http://www.ncbi.nlm.nih.gov/pubmed/16395542/	MFE and GC content	Dynamic programming
ptRNApred [45]	http://www.ptrnapred.org/	91 features including (1) 7 selected dinucleotide properties as well as their dinucleotide values, (2) 52 properties derived from the secondary structure, for example, the number of loops, and (3) 32 triplet element properties	Random forest and SVM
incRNA [46]	http://incrna.gersteinlab.org/	9 genomic features including 4 expression features, 3 sequence information, and 2 RNA structure features	Random forest