Multiscale RNA comparison based on RNA triple vector curve representation
 Ying Li^{1},
 Ming Duan^{2} and
 Yanchun Liang^{1}Email author
DOI: 10.1186/1471210513280
© Li et al.; licensee BioMed Central Ltd. 2012
Received: 10 May 2012
Accepted: 11 October 2012
Published: 30 October 2012
Abstract
Background
In recent years, the important functional roles of RNAs in biological processes have been repeatedly demonstrated. Computing the similarity between two RNAs contributes to better understanding the functional relationship between them. But due to the longrange correlations of RNA, many efficient methods of detecting protein similarity do not work well. In order to comprehensively understand the RNA’s function, the better similarity measure among RNAs should be designed to consider their structure features (base pairs). Current methods for RNA comparison could be generally classified into alignmentbased and alignmentfree.
Results
In this paper, we propose a novel waveletbased method based on RNA triple vector curve representation, named multiscale RNA comparison. Firstly, we designed a novel numerical representation of RNA secondary structure termed as RNA triple vectors curve (TVCurve). Secondly, we constructed a new similarity metric based on the wavelet decomposition of the TVCurve of RNA. Finally we also applied our algorithm to the classification of noncoding RNA and RNA mutation analysis. Furthermore, we compared the results to the two wellknown RNA comparison tools: RNAdistance and RNApdist. The results in this paper show the potentials of our method in RNA classification and RNA mutation analysis.
Conclusion
We provide a better visualization and analysis tool named TVCurve of RNA, especially for long RNA, which can characterize both sequence and structure features. Additionally, based on TVCurve representation of RNAs, a multiscale similarity measure for RNA comparison is proposed, which can capture the local and global difference between the information of sequence and structure of RNAs. Compared with the wellknown RNA comparison approaches, the proposed method is validated to be outstanding and effective in terms of noncoding RNA classification and RNA mutation analysis. From the numerical experiments, our proposed method can capture more efficient and subtle relationship of RNAs.
Keywords
RNA mutation Secondary structure RNAdistance RNApdist Multiscale RNA comparison RNA triple vector curveBackground
RNA once is considered as the fundamental information medium in central dogma of molecular biology. A number of studies have indicated that RNAs play a more active role and carry diverse functionalities in nature, including mediating the synthesis of proteins, regulating cellular activities, and exhibiting enzymelike catalysis and posttranscriptional activities. Furthermore, many recent discoveries have shown that the number and biological significance of functional RNAs has been underestimated. In living cells, RNAs do not remain in a linear form, which folds its secondary structure through base pairs including canonical bonds of AU and GC and wobble pair of GU. For understanding RNA's functionality, the alignment and similarity of RNA should consider not only the primary structure (sequence) but also the secondary structure (base pairs).
Numerous approaches were proposed to measure the similarity between RNA secondary structures, which can be broadly categorized into two classes: alignment based string or tree representation of RNA secondary structure, and comparison based some numerical representation without alignment.
Most studies usually adopt dynamic programming algorithms and tree models. Some are usually based on the alignment of a string representation of the secondary structures such as the dotbracket representation, in which a score function or a distance function to represent insertion, deletion and substitution of letters in the compared structures [1–4]. Sequences considered in alignment of RNA secondary structures are not only string sequences but also secondary structure. Different weights or different score functions are designed for unpaired nucleotides and paired nucleotides.
Others are almost based on alignment of a tree representation of the RNA secondary structure elements or the base pairing probability matrices [5–9]. Shapiro [5, 6] proposed various tree models used for representing RNA secondary structures without pseudoknots.
Each tree model offers a more or less detailed views of an RNA structure. Given the tree representations of two RNA secondary structure, one comparison way is based on the computation of the edit distance between the trees while the other focus on the alignment of the trees using the score of the alignment as a measure of the distance between the trees. Popular tools for optimal alignment of RNA secondary structures include RNAdistance [6] and RNAforester [8] etc. RNAdistance compares RNA secondary structures based on tree edit distance measure, while RNAforester computes the pairwise or multiple alignment of structures based on tree alignment measure. Hofacker [9] measured RNA secondary structures in terms of the base pairing probability matrices computed by McCaskill's partition function algorithm [10]. The popular tool based matrix of base pairing probabilities is RNApdist, which was implemented as part of the Vienna RNA package.
Because the above methods rely on dynamic programming algorithms, they are computationintensive even if the pseudoknots are ignored. For example, the Sankoff's algorithm [11] simultaneously allows the structure prediction and alignment problem with O(n^{4}) in memory and O(n^{6}) in time for two RNA sequences of lengthn. So these algorithms are still impractical for long RNA sequences. Recently some comparison algorithms without aligning them are proposed. Kin [12] gave a kernel method based on Stochastic Context Free Grammar (SCFG).
The graphical representations of biosequences (protein, DNA and RNA) could be out of the mainstream but a new research view and tool to understand and analyze such biosequences. M.Randic [13] reviewed the sufficient materials on related topics of graphical representations of protein, DNA and the secondary structure of RNA. Inspired by several graphical representations of DNA sequences [14–18], some researchers have proposed 2D, 3D or 4D graphical approaches for the representations of RNA secondary structure and then derive some numerical invariants and different graphical measures from graphs to compare RNA secondary structures [19–32].
In [19–29], eight symbols of the unpaired bases A, C, G, U and paired bases A′, C′, G′, U′ were used to code RNA secondary structures as graphical representations. In [31], the representations of eight letters have been demonstrated to be approximate and have some loss of information. In [32], 12 symbols have been used to represent RNA secondary structure without loss of information, in which the key is to discriminate between the first and the second base of a hydrogen bond for the paired bases.
In this paper, motivated by DV curve representation of DNA sequences [33, 34], we propose a novel triple vector curve representation of RNA secondary structure. With this novel representation, a new RNA secondary structure similarity measure based on wavelet analysis is designed, which can simultaneously focus on the local structure and global structure. To evaluate our algorithms, we take the classification of noncoding RNA and RNA mutation as examples to compare to the two popular tools of RNAdistance and RNApdist.
Results and discussion
Similarities/dissimilarities among noncoding RNA from different families
We performed the experiments on 100 RNA sequences to test the ability to distinguish noncoding RNA families. We randomly chose 25 sequences from each of the four RNA classes (5S rRNA, miRNA, RNaseP arch and tRNA) in RFAM database.
Firstly, the secondary structures of the 100 RNA sequences are predicted by the Vienna RNA folding prediction package. Secondly, their characteristic representations are constructed according to the primary sequence and the predicted secondary structure. Thirdly, the TVCurves can be obtained based on their characteristic representations. Then we computed the similarity between any two RNA among these 100 RNA sequences by the proposed multiscale similarity measure algorithm based on TVCurve. Furthermore, all the similarity values are arranged into a similarity matrix. For validation of our algorithm, we computed the distance matrixes using RNApdist and RNAdistance tools respectively.
where $\stackrel{\u2015}{X}$ and $\stackrel{\u2015}{Y}$ denote the means of the matrices of X and Y. The larger absolute value of H indicates the better coherence between the similarity matrix X and the ground matrix Y. The value of H can be used to estimate the quality of the similarity measure.
The Hubert statistic comparison for different similarity matrixes for 100 noncoding RNAs
Method  RNApdist  RNAdistance  Multiscale similarity based on TVCurve 

Hubert statistic  0.4095  0.1156  0.7205 
However, the Hubert's statistic of our proposed multiscale similarity measure based on our algorithm is 0.7205. Obviously, our similarity measure is more closer to the real data compared with RNApdist and RNAdistance.
Obviously, compared Additional file 1: Figure S1 and Figure S2 with Figure 1, the phylogenetic tree based on our proposed measure presents clearly four branches. The four branches of Figure 1 can be regarded as the classification of the 100 RNA sequences, where branch 1 with 23 RNaseP_archs, branch 2 with 22 tRNAs and 2 RNaseP_archs, branch 3 with 25 miRNA and 2 tRNA, and branch 4 with 25 5S_rRNAs and 1 tRNAs. It is easy to obtain the false percentage is 5%. Moreover, the 2 RNaseP_archs in branch 2 and 2 tRNA in branch 3 are both isolated from the 22 tRNAs and 25 miRNA respectively. The distinguished performance of our proposed method is better than the popular RNA comparison RNApdist and RNAdistance tools.
Similarities/dissimilarities among the RNA secondary structures of nine virus
The similarity matrix for the secondary structures at the 3'terminus belonging to nine viruses of Figure 2 by multiscale RNA comparison based on RNA triple vector curve representation.
Species  ALMV  CiLRV  TSV  CVV  APMV  LRMV  PDV  EMV  AVII 

ALMV  1.0000  0.2596  0.2300  0.1281  0.1638  0.2606  0.4545  0.2688  0.3770 
CiLRV  1.0000  0.3259  0.4983  0.2678  0.5929  0.2007  0.4241  0.4337  
TSV  1.0000  0.3828  0.2869  0.2888  0.3054  0.1652  0.1443  
CVV  1.0000  0.3947  0.6029  0.2755  0.3217  0.5566  
APMV  1.0000  0.1912  0.7407  0.3245  0.1886  
LRMV  1.0000  0.1734  0.4963  0.6387  
PDV  1.0000  0.3033  0.1187  
EMV  1.0000  0.4248  
AVII  1.0000 
Observing Table 2 and Figure 4, we find the most similar species pairs are (PDV,APMV) and (AVII, LRMV), and the next similar species pairs are (ALMV,PDV),(ALMV,APMV), (AVII, CVV) and (CVV,LRMV). The results are analogous to the difference of the secondary structures in Figure 2, which show that our approach also present the better performance for similar secondary structure comparison.
RNA mutation analysis
Additionally, in order to further validate the efficiency of our method, we test the 21 rRNA fragments of the thermus thermophilus from Ribosomal dataset in [37] compared with RNAdistance and RNApdist. The labels and sequences are listed in Table S1 (See Additional file 2: Table S1). In Table S2 (See Additional file 2: Table S2), the most significant mutation position and type are listed for RNAmscTVcurve (RNA multiscale RNA comparison based on RNA triple vector curve), RNAdistance and RNApdist. Out of the 21 RNA sequences in the data set, 3 fragments (E_(68),A_(588–651),A_(1113–1187), A_(240–286)) produced the same most significant mutation as RNAdistance and RNApdist. Our proposed structure for the rest fragments are more different than the structure with the largest RNAdistance and RNApdist but it is nonobvious to determine which one of them is more significant. Both of the mutations alter the structure with respect to the original structure. The results in Figure S3 (See Additional file 1: Figure S3) provide further evidence that our method can capture more significant and subtle structure mutation.
Conclusion
In this paper, we provide a better visualization and analysis tool TVCurve for RNA to indicate the information of sequence and secondary structure especially for long RNA. Additionally, based on TVCurves representation of RNA, a multiscale similarity measure for RNA comparison is proposed, which can capture the local and global difference between the information of sequence and structure of RNA. Compared with the popular RNA comparison approaches, the proposed method is evaluated to be outstanding and effective. But as we know, the native secondary structure of a RNA is often a suboptimal structure not the predicted structure with minimum free energy (MFE) due to limitations of thermodynamic models. The structural similarity measurement using multiple predicted suboptimal structures is still a challenge. In the further research, we will focus on how to measure the structural similarity to integrate multiple structures with different energy levels.
Method
The TVCurve representation of RNA secondary structure
In this section, we describe the construction of TVCurve of the secondary structure of RNA. Firstly, we give the characteristic representation of RNA based on the primary and secondary structure of RNA.
The characteristic representation of RNA secondary structure
In [15], Liao proposed a characteristic representation of RNA secondary structure, which both include the primary structure and secondary structure. This representation was based on eight symbols, the four A, C, G, U for the four nucleotide bases (adenine, cytosine, guanine and uracil, respectively) and four A', C', G', U' for the same bases if paired by hydrogen bonds. In the primary structure of RNA, let A', U', G' and C' denote A, U, G and C in the base pair AU, GC or GU, respectively. A characteristic sequence of the secondary structure can be obtained. The RNA secondary structure is predicted by the Vienna RNA folding prediction package [38].
Combining the information of the sequence and secondary structure, we give the corresponding characteristic sequence of the secondary structure of tRNA ( U48228.1/7166 ) in the following:>tRNA (U48228.1/7166)
CAAU'C'U'UAA'CG'A'U'G'G'AUG'U'C'U'U'GG'U'
U'CC'UAUAG'CG'A'U'GA'A'G'G'C'C'G'C'A'G'C'A'AAGU'
G'C'GAUAU'G'C'A'AU'GAAAA'AU'G'C'A'AUUACU'G'U'
G'AAU'C'A'U'C'AG'A'A'U'G'C'UGAA'U'G'U'AAA'CUAU
AC'C'A'U'A'UUUACCCUU'A'U'G'G'GCAAAU'UAA'C'G'
U'GG'U'A'U'U'C'CU'ACA'G'A'AA.
Construction of TVCurve
For a given TVCurve of RNA, we can retrieve its characteristic representation from equation (2).
 (1).
TVCurve extends 3 units along Xaxis to represent each unpaired nucleotide (A, T, C G) and paired nucleotide (A', T', C' G').
 (2).
From TVCurve, one can immediately grasp the information about RNA sequence and structure information. From a given TVCurve, we can obtain its unique sequence and secondary structure representation. Moreover, for a given RNA sequence and structure, there is a unique TVCurve representation. The correspondence between TVCurves and the RNA information of sequences and secondary structures is one to one and no loss of information. If one wants to know whether the ith nucleotide in RNA sequence is paired, only need to examine the difference between the values at (3i2) and (3i3) of TVCurve. If y_{3i − 2} − y_{3i − 3} = 1, the ith nucleotide is paired. If y_{3i − 2} − y_{3i − 3} = − 1, the ith nucleotide is unpaired.
 (3).
The Xaxis end point x_{ end } of the TVCurve indicates the length of RNA sequence n, i.e. n = x_{ end }/3.
Multiscale similarity measure based on TVCurves
In this section, based on TVCurves of RNA, we propose a multiscale similarity measure for RNA comparison in terms of the multiscale property of wavelet transform.
We estimate RNA similarity using the weighted correlations in the wavelet domains at the different scales. The main characteristics of wavelet transforms are timefrequency localization and multiresolution property. Wavelet can capture the global and local property of a signal synchronously and can focus on the any detail of a signal. In this sense wavelets are referred to as a mathematical microscope. In the following, we briefly introduce the discrete wavelet transform [39, 40].
The wavelet transform relies on the wavelet function ψ(x) and the scaling function ∅ ϕ(x), which
Where {h_{ n }} is a lowpass filter (scaling filter).
Where {g_{ n }} is a highpass filter (wavelet filter)
Given a signal s with length N, the wavelet transform consists of log_{ 2 }N levels at most. The wavelet decomposition of the signal s analyzed at level one is provided with two sets of coefficients: approximation coefficients cA_{ 1 }, and detail coefficients cD_{1}. cA_{1} is obtained by convolving s with the lowpass filter and then is downsampled (keep the even index elements) for approximation, and cD_{ 1 } is also obtained by the highpass filter and then is downsampled for detail.
And if {h_{ l }}_{ l } and {g_{ l }}_{ l } are orthogonal, there is g_{ l } = (−1)^{ l }h_{1 − l}. While in the biorthogonal condition there are four filters (two group filters): decomposition filters {h_{ l }}_{ l }, {g_{ l }}_{ l }, reconstruction filters ${\left\{{\tilde{h}}_{l}\right\}}_{l},{\left\{{\tilde{g}}_{l}\right\}}_{l}$, where ${g}_{l}={\left(1\right)}^{l}{\tilde{h}}_{1l}$ and ${\tilde{g}}_{l}={\left(1\right)}^{l}{h}_{1l}$.
Abbreviations
 TVcurve:

Triple vector curve
 RNAmscTVcurve:

RNA multiscale comparison based on Triple vector curve.
Declarations
Acknowledgements
This research was supported by the National Natural Science Foundation of China (11001106, 61073075 and 61272207), and the ScienceTechnology Development Project from Jilin Province of China (20120730). The authors would like to thank the editor and two anonymous reviewers for their numerous helpful suggestions and comments for this manuscript.
Authors’ Affiliations
References
 Gardner PP, Wilm A, Washietl S: A benchmark of multiple sequence alignment programs upon structural RNAs. Nucleic Acids Res 2005, 33(8):2433–2439. 10.1093/nar/gki541PubMed CentralView ArticlePubMedGoogle Scholar
 Dowell RD, Eddy SR: Efficient pairwise RNA structure prediction and alignment using sequence alignment constraints. BMC Bioinforma 2006, 7: 400. 10.1186/147121057400View ArticleGoogle Scholar
 Konings DA, Hogeweg P: Pattern analysis of RNA secondary structure similarity and consensus of minimalenergy folding. J Mol Biol 1989, 207(3):597–614. 10.1016/00222836(89)904683View ArticlePubMedGoogle Scholar
 Havgaard JH, Torarinsson E, Gorodkin J: Fast pairwise structural RNA alignments by pruning of the dynamical programming matrix. PLoS Comput Biol 2007, 3(10):1896–1908.View ArticlePubMedGoogle Scholar
 Shapiro BA: An algorithm for comparing multiple RNA secondary structures. Computer applications in the biosciences: CABIOS 1988, 4(3):387–393.PubMedGoogle Scholar
 Shapiro BA, Zhang KZ: Comparing multiple RNA secondary structures using tree comparisons. Computer applications in the biosciences: CABIOS 1990, 6(4):309–318.PubMedGoogle Scholar
 Allali J, Sagot MF: A new distance for high level RNA secondary structure comparison. Ieee Acm T Comput Bi 2005, 2(1):3–14.Google Scholar
 Hochsmann M, Toller T, Giegerich R, Kurtz S: Local similarity in RNA secondary structures. Proceedings /IEEE Computer Society Bioinformatics Conference IEEE Computer Society Bioinformatics Conference 2003, 2: 159–168.PubMedGoogle Scholar
 Hofacker IL FW, Stadler PF, Bonhoeffer LS, Tacker M, Schuster P: Fast folding and comparison of RNA secondary structures. Monatshefte für Chemie 1994, 125: 167–188. 10.1007/BF00818163View ArticleGoogle Scholar
 McCaskill JS: The equilibrium partition function and base pair binding probabilities for RNA secondary structure. Biopolymers 1990, 29(6–7):1105–1119.View ArticlePubMedGoogle Scholar
 Sankoff D: Simultaneous solution of the RNA folding alignment and protosequence problems. SIAM J Appl Math 1985, 45: 810–825. 10.1137/0145048View ArticleGoogle Scholar
 Kin T, Tsuda K, Asai K: Marginalized kernels for RNA sequence data analysis. Genome informatics International Conference on Genome Informatics 2002, 13: 112–122.PubMedGoogle Scholar
 Randic M, Zupan J, Balaban AT, VikicTopic D, Plavsic D: Graphical Representation of Proteins. Chem Rev 2011, 111(2):790–862. 10.1021/cr800198jView ArticlePubMedGoogle Scholar
 Randic M, Basak SC: Characterization of DNA primary sequences based on the average distances between bases. J Chem Inf Comp Sci 2001, 41(3):561–568. 10.1021/ci0000981View ArticleGoogle Scholar
 Randic MVM, Lers N, Plavsic D: Analysis of similarity/dissimilarity of DNA sequences based on novel 2D graphical representation. Chem Phys Lett 2003, 371: 202–207. 10.1016/S00092614(03)002446View ArticleGoogle Scholar
 Guo XF, Nandy A: Numerical characterization of DNA sequences in a 2D graphical representation scheme of low degeneracy. Chem Phys Lett 2003, 369(3–4):361–366.View ArticleGoogle Scholar
 Zupan J, Randic M: Algorithm for coding DNA sequences into "spectrumlike" and "zigzag" representations. J Chem Inf Model 2005, 45(2):309–313. 10.1021/ci040104jView ArticlePubMedGoogle Scholar
 Liao B, Wang TM: 3D graphical representation of DNA sequences and their numerical characterization. J Mol StrucTheochem 2004, 681(1–3):209–212.View ArticleGoogle Scholar
 Liao B, Wang TM: A 3D graphical representation of RNA secondary structures. J Biomol Struct Dyn 2004, 21(6):827–832. 10.1080/07391102.2004.10506972View ArticlePubMedGoogle Scholar
 Jiaquan Zhan BL, Yusen Z: Numerical characterization of RNA secondary structure. Internet Electronic Conference of Molecular Design 2003., 2003: November 23 – December 6, ttp://biochempress.com/Files/IECMD_2004/IECMD_2004_018.pdf November 23 – December 6,Google Scholar
 Bai FL, Zhu W, Wang TM: Analysis of similarity between RNA secondary structures. Chem Phys Lett 2005, 408(4–6):258–263.View ArticleGoogle Scholar
 Feng J, Wang TM: A 3D graphical representation of RNA secondary structures based on chaos game representation. Chem Phys Lett 2008, 454(4–6):355–361.View ArticleGoogle Scholar
 Liu LW, Wang TM: On 3D graphical representation of RNA secondary structures and their applications. J Math Chem 2007, 42(3):595–602. 10.1007/s1091000691354View ArticleGoogle Scholar
 Yao YH, Nan XY, Wang TM: A class of 2D graphical representations of RNA secondary structures and the analysis of similarity based on them. J Comput Chem 2005, 26(13):1339–1346. 10.1002/jcc.20271View ArticlePubMedGoogle Scholar
 Yao YH, Liao B, Wang TM: A 2D graphical representation of RNA secondary structures and the analysis of similarity/dissimilarity based on it. J Mol StrucTheochem 2005, 755(1–3):131–136.View ArticleGoogle Scholar
 Li C, Xing LL, Wang X: Analysis of similarity of RNA secondary structures based on a 2D graphical representation. Chem Phys Lett 2008, 458(1–3):249–252.View ArticleGoogle Scholar
 Zhu W, Liao B, Ding KQ: A condensed 3D graphical representation of RNA secondary structures. J Mol StrucTheochem 2005, 757(1–3):193–198.View ArticleGoogle Scholar
 Zhang Y, Qiu JQ, Su LQ: Comparing RNA secondary structures based on 2D graphical representation. Chem Phys Lett 2008, 458(1–3):180–185.View ArticleGoogle Scholar
 Liao B, Zhu W, Li PC: On a fourdimensional representation of RNA secondary structures. J Math Chem 2007, 42(4):1015–1022. 10.1007/s109100069156zView ArticleGoogle Scholar
 Zhang YS: On 3D graphical representation of RNA secondary structure. MatchCommun Math Co 2007, 57(1):157–168.Google Scholar
 Liao B, Chen W, Sun X, Zhu W: A binary coding method of RNA secondary structure and its application. J Comput Chem 2009, 30(14):2205–2212.PubMedGoogle Scholar
 Randic M, Plavsic D: Novel spectral representation of RNA secondary structure without loss of information. Chem Phys Lett 2009, 476(4–6):277–280.View ArticleGoogle Scholar
 Zhang ZJ: DVCurve: a novel intuitive tool for visualizing and analyzing DNA sequences. Bioinformatics 2009, 25(9):1112–1117. 10.1093/bioinformatics/btp130View ArticlePubMedGoogle Scholar
 Randic M, Vracko M, Novic M, Plavsic D: SpectrumLike Graphical Representation of RNA Secondary Structure. Int J Quantum Chem 2009, 109(13):2982–2995. 10.1002/qua.22175View ArticleGoogle Scholar
 Tseng VS, Kao CP: Efficiently mining gene expression data via a novel parameterless clustering method. Ieee Acm T Comput Bi 2005, 2(4):355–365.Google Scholar
 Ronquist F: Inferring phylogenies. Science 2004, 303(5659):767–768. 10.1126/science.1094283View ArticleGoogle Scholar
 Ivry T, Michal S, Avihoo A, Sapiro G, Barash D: An image processing approach to computing distances between RNA secondary structures dot plots. Algorithm Mol Biol 2009., 4:Google Scholar
 Wuchty S, Fontana W, Hofacker IL, Schuster P: Complete suboptimal folding of RNA and the stability of secondary structures. Biopolymers 1999, 49(2):145–165. 10.1002/(SICI)10970282(199902)49:2<145::AIDBIP4>3.0.CO;2GView ArticlePubMedGoogle Scholar
 Mallat SG: A theory for multiresolution signal decomposition  the wavelet representation. Ieee T Pattern Anal 1989, 11(7):674–693. 10.1109/34.192463View ArticleGoogle Scholar
 Daubechies I: Ten lectures on wavelets. Philadelphia, PA, USA: Society for Industrial and Applied Mathematics; 1992.View ArticleGoogle Scholar
Copyright
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.