Skip to main content

Table 1 Overview of ncRNA search tools assessed in this study

From: Considerations in the identification of functional RNA structural elements in genomic alignments

Tool

Description

URL

Ref

MSARI

Developed for use with 10–15 multiple alignments, looks for compensatory mutations near bases predicted to pair in secondary structure. Uses RNAfold [33] to predict which bases pair and analyzes neighbouring nucleotide pairs in a 7 nt window for presence of compensatory mutations.

http://theory.csail.mit.edu/MSARi/

[12]

RNAz

Uses two variables to assess ncRNA potential in two or more sequence alignments: a z-score and an SCI score. The z-score is a measure of the thermodynamic stability of the reference sequence relative to shuffled variants of the same sequence. RNAz samples the z-scores using machine learning from pre-computed z-scores covering a range of sequence lengths and sequence compositions, rather than re-computing z-scores for each sequence. The SCI (structure conservation index) is an indirect measure of structural similarity among the individual sequences in the alignment based on their individual folding energies compared to the folding energy of their consensus structure. This term also incorporates a covariance factor, rewarding compensatory mutations. RNAz uses machine learning to classify alignments as ncRNAs on a combination of their z- and SCI scores. Scores reflect how far the alignment is from the ncRNA line of separation in an SCI-z plane.

http://www.tbi.univie.ac.at/~wash/RNAz/

[10]

ddbRNA

ddbRNA scans for compensatory mutations in conserved stem loops found within two or more sequence alignments. It counts the number of compensatory mutations in all possible hairpins of the alignment and compares that count to a distribution of counts from shuffled variants of the alignment. The score output is the number of standard deviations above the mean count of the shuffled variants (negative scores indicate more compensatory mutations in the shuffled variants).

Contact authors.

[11]

QRNA

QRNA is a probabilistic algorithm that predicts pairwise sequence alignments as belonging to one of three classes: coding (COD), noncoding (RNA), or other (OTH). The RNA model incorporates two components: a stochastic-context free grammar (SCFG) to estimate a distribution of probabilities over potential structures and a pair-Hidden Markov Model to predict the probability that the structures were evolutionarily selected on the basis of a compensatory substitution pattern in the alignment. The logoddspostRNA score reflects these two components and is used as the primary discriminant score in this study.

http://selab.wustl.edu/cgi-bin/selab.pl?mode=software

[9]

Evofold

Evofold computes the probability that the observed sequence alignment was generated under selection for a functional RNA (structural) versus evolutionary divergence of non-structural sequence. It uses the traditional SCFG algorithm CYK without explicitly defining emission probabilities over observed aligned bases, but rather uses an evolutionary model (Felsenstein) to compute the probability of the column given a phylogenetic tree. The functional RNA (fRNA) model comprises both a structural and non-structural components. The structural component computes the probability that two columns pair (i.e. occur in stems) whereas the non-structural component computes the probability of observing single-nucleotide columns.

http://www.cbse.ucsc.edu/~jsp/EvoFold/

[13]

zMFOLD

This is in an iterative script that utilizes hybrid-ss-min (MFOLD variant with updated handling of partition function calculations) [34] and shuffle-pair.pl to generate z-scores that are indicative of selection for thermodynamic stability over the expected stability of a random sequence with an identical dinucleotide composition. The z-score is the number of standard deviations that the actual stability is below the mean stability of 100 shuffles (using shuffle-pair.pl). Higher z-scores indicate apparent selection for structural stability. The thermodynamic stability is calculated only for the reference strand, while the alignment constrains the shuffling.

http://www.bioinfo.rpi.edu/applications/

hybrid/OligoArrayAux.php

http://hugheslab.med.utoronto.ca/Babak/

tools/zMFOLD.pl

(perl script that calls hybrid-ss-min and shuffle-pair.pl)

Additional file 4

[34]

zRNAfold

Same as zMFOLD except RNAfold [35] is used to predict thermodynamic stabilities.

http://www.tbi.univie.ac.at/~ivo/RNA/

[35]

zRNAfold (Dual)

z-scores are computed with zRNAfold for both strands in the alignment and added for a final z-score.

http://www.tbi.univie.ac.at/~ivo/RNA/

[35]

Alifoldz

Predicts selection for structure by comparing the minimum free energy of a consensus structure to shuffles of the alignment. Scores are multiplied by -1 to scale with other algorithms where higher scores indicate structure.

http://www.tbi.univie.ac.at/papers/SUPPLEMENTS/Alifoldz/

[18]