Bioinformatics Advance Access originally published online on June 30, 2005
Bioinformatics 2005 21(18):3610-3614; doi:10.1093/bioinformatics/bti562
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
MicroRNA identification based on sequence and structure alignment
MOE Key Laboratory of Bioinformatics, Department of Automation, Tsinghua University Beijing 100084, China
*To whom correspondence should be addressed.
 |
Abstract |
|---|
 TOP Abstract INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
Motivation: MicroRNAs (miRNA) are 
22 nt long non-coding RNAs that are derived from larger hairpin RNA precursors and play important regulatory roles in both animals and plants. The short length of the miRNA sequences and relatively low conservation of pre-miRNA sequences restrict the conventional sequence-alignment-based methods to finding only relatively close homologs. On the other hand, it has been reported that miRNA genes are more conserved in the secondary structure rather than in primary sequences. Therefore, secondary structural features should be more fully exploited in the homologue search for new miRNA genes.
Results: In this paper, we present a novel genome-wide computational approach to detect miRNAs in animals based on both sequence and structure alignment. Experiments show this approach has higher sensitivity and comparable specificity than other reported homologue searching methods. We applied this method on Anopheles gambiae and detected 59 new miRNA genes.
Availability: This program is available at http://bioinfo.au.tsinghua.edu.cn/miralign
Contact: daulyd{at}tsinghua.edu.cn
Supplementary information: Supplementary information is available at http://bioinfo.au.tsinghua.edu.cn/miralign/supplementary.htm
|
INTRODUCTION |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
MicroRNAs (miRNAs) are
22 nt long endogenous non-coding RNAs that play important regulatory roles in diverse organisms (Bartel and Bartel, 2003; Bartel, 2004; He and Hannon, 2004; Lai, 2003). In animal cells, miRNA genes are first transcribed as long pri-miRNAs (Lee et al., 2002) and processed to 70 nt precursors (pre-miRNA) with stem–loop structure by the RNase III enzyme Drosha (Lee et al., 2003). Then, another RNase III enzyme Dicer (Grishok et al., 2001; Hutvagner et al., 2001; Ketting et al., 2001) cuts the pre-miRNAs to release the 22 nt mature miRNAs (Lee et al., 2002; Lee et al., 2003). Finally, RNA-induced silencing complexes (Hammond et al., 2000) are formed to regulate the expression of target genes via complementary base pair interactions. In plants, the maturation process of miRNAs is similar to that in animals, but the length of the pre-miRNAs are more variable and their structures are more complex (Bartel, 2004). Since the first miRNA was discovered in Caenorhabditis elegans (Lee et al., 1993) hundreds of miRNAs have been cloned in many organisms (Lagos-Quintana et al., 2001; Lau et al., 2001; Lee and Ambros, 2001). However, only abundant miRNA genes can be easily detected by PCR or northern blot due to limitations of the techniques. For finding those low-expression or tissue-specific miRNA genes computational prediction provides an efficient strategy(Bartel, 2004).
Up to now, several computational approaches have been reported to identify miRNAs. Comparative genomics was adopted to find entirely novel miRNA families in specific animals and plants (Bonnet et al., 2004; Jones-Rhoades and Bartel, 2004; Lai et al., 2003; Lim et al., 2003a; Lim et al., 2003b; Ohler et al., 2004; Wang et al., 2004). Homologue search was used to reveal orthologs and paralogs of known miRNAs (Lagos-Quintana et al., 2001; Lau et al., 2001; Lee and Ambros, 2001; Maher et al., 2004; Pasquinelli et al., 2000; Weber, 2005). Because the mature miRNA sequences are short (22 nt), current sequence alignment tools like BLAST (Altschul et al., 1990) can only find the nearly-perfect matches due to the large number of irrelevant hits. The 70 nt pre-miRNA sequences have also been used for homologue search. However, compared with miRNA and miRNA* (the fragments on the opposite arm of the hairpin) (Lau et al., 2001) the other parts of the precursor sequence are less conserved. Therefore, sequence alignment alone may fail to detect the distant homologs that diverge in sequence but keep conservation in structure (Legendre et al., 2004). So, more sensitive approaches that take the advantage of both sequence and structure conservation are needed. Recently, an ERPIN (Gautheret and Lambert, 2001) search strategy was reported (Legendre et al., 2004), which used profiles to capture both sequence and structure information of animal miRNA precursors. This method increased the number of found novel miRNA candidates by 17% compared with BLAST search, but the construction of profile makes it only applicable to miRNA families with sufficient known samples.
In this paper, we present a novel computational approach named miRAlign that detects new miRNAs based on both sequence and structure alignment. Two main characters make miRAlign distinct from existing homologue search methods: firstly, to be able to find distant homologs, miRAlign requires neither the sequence conservation of the whole pre-miRNA sequence nor the nearly-perfect match of the 22 nt mature part, but just assumes relatively loose conservation of the mature miRNA sequence. Secondly, more properties of miRNA structure conservation are considered. And unlike profile search methods, which need relatively large family members to construct the profile, miRAlign introduces a structure alignment strategy and can use each single miRNA as a query to do homology search. In our experiments, miRAlign outperforms conventional BLAST search and ERPIN search by higher sensitivity and comparable specificity. Using this method, more distant miRNA homologs or orthologs can be revealed.
|
MATERIALS AND METHODS |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
miRNA reference sets and genomic sequences
The miRNAs and their precursor sequences were downloaded from the MicroRNA Registry (Ambros et al., 2003; Griffiths-Jones, 2004) release 5.0 (http://www.sanger.ac.uk/Software/Rfam/mirna/index.shtml). This set contains 1298 miRNAs (generated by 1345 pre-miRNAs) from 12 species (C.elegans, Caenorhabditis briggsae, Drosophila melanogaster, Drosophila pseudoobscura, Danio rerio, Gallus gallus, Homo sapiens, Mus musculus, Rattus norvegicus, Epstein Barr virus, Arabidopsis thaliana and Oryza sativa), of which 1054 are from animals. Most of these miRNAs were identified or verified by experiments, and others were computationally predicted as their close homologs. These 1054 animal miRNAs and their precursors (1104 pre-miRNAs) composed our raw training set Train_All. In order to illustrate miRAlign performance on C.briggsae genomic sequences, two sub-training sets were constructed. Train_Sub_1 consists of all the known animal miRNAs except for those from C.briggsae. It contains 976 miRNAs and 1025 pre-miRNA sequences. Train_Sub_2 was constructed by removing both the C.briggsae and C.elegans miRNAs from Train_All, leaving 859 miRNAs and 909 precursors.
Genomic sequences of six species were used in this work, including Anopheles gambiae (MOZ2a, http://www.ensembl.org/), C.briggsae (cb25.agp8, http://www.sanger.ac.uk/Projects/C_briggsae), C.elegans (WormBase release WS 140, www.wormbase.org), D.melanogaster (release 4, http://www.fruitfly.org/annot/release4.html), D.pseudoobscura (freeze1, http://www.hgsc.bcm.tmc.edu/projects/drosophila/) and G.gallus (WASHUC1, http://www.genome.wustl.edu/projects/chicken).
Extraction of miRNA candidate sequences (preprocessing)
Several preprocessing steps are taken to draw miRNA candidate sequences from the genome. Firstly, all the known pre-miRNAs in the training set are used as queries to BLAST search against the genome with a sensitive BLAST parameter setting (word-length 7 and E-value cutoff 10). Next, sequence segments of the potential regions are cut from the genome with 70 nt flanking sequences to each end and scanned by a 100 nt-sliding window with a step of 10 nt. The sequences overlapping with repeat sequences are discarded and the remains are treated as miRNA candidates to be scored by miRAlign.
miRAlign
miRAlign is designed to find new miRNAs using both sequence information and structural characteristics of known miRNAs. For a given 70 nt candidate miRNA precursor sequence, miRAlign assigns it a similarity score by following five steps.
(1) Secondary structure prediction. Both the candidate sequence and its reverse complementary are analyzed by miRAlign. The secondary structures of both strands of the candidate are predicted by RNAfold (Hofacker et al., 1994) using minimum free energy (MFE) rule (Zuker and Stiegler, 1981). Only the strands with MFE lower than –20 kcal/mol are kept for further analysis.
(2) Pairwise sequence alignment. The strands of the candidate sequences that pass Step 1 are pairwisely aligned to all the 22 nt known miRNA sequences in the training set. Sequence similarity score (mature_seq_sim) between the candidate and each known mature miRNAs is calculated by CLUSTALW (Thompson et al., 1994). The candidate-to-known miRNA pairs are kept as potential homologue pairs for further analysis only if their mature_seq_sim exceed a user-defined threshold min_seq_sim. Here, min_seq_sim = 70 is selected as the default parameter for miRAlign in our experiments. More than 98.1% of the known animal miRNA homologs have sequence similarity higher than this value (Supplementary Figure S1). For each of those potential homologue pairs, the 22 nt sub-sequence on the candidate that aligns to the known miRNA is regarded as potential miRNA.
(3) miRNA's position on the stem–loop structure. Three properties for the 22 nt miRNA's position on the stem–loop structure derived from the miRNA reference set are considered by miRAlign for each of the potential homologue pairs: (a) the 22 nt potential miRNA sequence should not locate on the terminal loop of the hairpin structure; (b) potential miRNA should locate on the same arm of the stem–loop structure as its known homologs and (c) the position of the potential miRNA sequence on the stem–loop structure should not differ too much from its known homologs.
The position difference of the mature miRNAs (or potential mature miRNAs) on the stem–loop structure between precursor A and B is calculated by
 |
22 nt miRNA and miRNA* (Fig. 1). A 
_len value close to zero indicates that there are few deletions or insertions occurred between the miRNA homologs. In our experiments, a non-stringent _len cutoff 15 was used as the default parameter. Over 97.5% of the known animal miRNA homologue pairs have _len less than this value (Supplementary Figure S2).
|
(4) RNA secondary structure alignment. To measure the secondary structure conservation of the potential homologue pairs, RNAforester (Hochsmann et al., 2003) is used to compute pairwise structure alignment. This tool implements a tree alignment model. It can calculate local structure alignment and give a similarity measure of two structures. Here, the parameter ‘-RIBOSUM’ is used, and RNAforester calculates structure alignment considering both sequence and structure information by using the base and base pair substitution matrix RIBOSUM85-60 as described in Klein and Eddy (2003). Using this log-odd position independent substitution matrix, this tool can use a single sequence as a query to align against the target sequence (while profile-based scoring scheme like ERPIN needs large family members to construct the profile and compute position dependent log-odds scores).
The raw structure alignment score 
local computed by RNAforester is the summation of all base (base pair) match (insertion, detention, etc.) scores in the alignment. It cannot be used directly to measure the similarity for the structures with different sizes. We define the normalized similarity score of structure C and m as
 |
local(C,m) denotes the raw local alignment score of two structures C and m, and (m,m) is the self-alignment score of m. After the normalization by (m,m), str_sim ranges from 0 to 100. The higher this value, the more similar are the two structures.
(5) Total similarity score. After aligning all the potential homologue pairs, a total similarity score (tss) is assigned to the candidate sequence. This tss is defined as
 |
An overview description of the miRAlign procedures is shown in Figure 2.
|
|
RESULTS AND DISCUSSION |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
Application on C.briggsae
MicroRNA Registry 5.0 contains 79 C.briggsaes miRNAs, all of which were predicted by the homologous search. So we first applied our method on C.briggsaes data with the training set Train_Sub_1 to check the sensitivity and specificity of miRAlign. Then Train_Sub_2 was used as the training set to investigate the ability of miRAlign to find new miRNAs between distantly related species.
Detection of miRNA homologs
Using training set Train_Sub_1, which contains all the animal miRNAs except for those from C.briggsae, 96 500 candidate sequences passed the preprocessing step and all of them were scored by miRAlign. Candidates belonging to same regions on the genome were merged and represented by the one with the highest tss score. In total, 1050 non-overlapping candidates got none-zero tss, including 74 miRNAs among the 79 reported C.briggsae miRNA genes (Fig. 3A). With a tss cutoff of 35, 90 putative miRNA sequences were detected, and a sensitivity of 89.9% was achieved by finding 73 copies of 71 known C.briggsae miRNA genes. Supplementary Table S1 lists the novel putative miRNAs with tss >35. Eight genuine C.briggsae miRNA genes with tss cutoff 35 were not identified by miRAlign. Four of them were missed during the BLAST search in the preprocessing step; one's mature part overlaps with the hairpin-loop; and the other three got tss <35. For the four ‘BLAST-missing’ miRNAs, we used their reported precursor sequences directly as the input of miRAlign and three of them got tss >35.
|
Identification of miRNAs in distantly related species
All C.briggsae miRNAs in MicroRNA Registry 5.0 were predicted based on close homologs to verified miRNAs from C.elegans. Obviously, the C.elegans miRNAs in the training set greatly contribute to the homologue search results. In order to investigate the ability of miRAlign to identify miRNAs in distantly related species, we performed the experiment described above by using the training set Train_Sub_2, in which both the C.briggsae and C.elegans miRNAs are excluded. There were 18 putative miRNAs that got tss >35, but only 8 known C.briggsae miRNAs genes were included (Table 1). We investigated these known miRNAs that failed to be detected in this experiment and found that most of them were lost in the preprocessing step. If we directly use miRAlign to score the known C.briggsae pre-miRNAs with a sensitive parameter mature_seq_sim 60, 20 known miRNAs could get tss >35. This suggests that miRAlign can be used to find miRNAs in relatively distant related species with the help of more sensitive sequence alignment approaches in the preprocessing step. This character will facilitate the extension of known miRNA set from the current limited number of species to a wide range. However, we also noticed that there still remain some miRNA homologs failed to be detected by miRAlign because of their greater divergence in either primary sequences or secondary structures.
|
tss cutoff selection
As shown in Figure 3A, we may detect more putative miRNAs by using a lower tss cutoff. However, the increase in sensitivity will decrease the specificity. So it is essential to find out a tss cutoff that has reasonable selectivity and satisfactory sensitivity. Because the number of false positives is inaccessible directly without experimental verification, we estimated it based on the miRAlign hits on 10 randomly shuffled C.briggsae genomes using the training set Train_Sub_1. When choosing a tss of 45 as a cutoff, 86.1% sensitivity is achieved and no false positive hits are found. When the cutoff decreases to 35, the sensitivity increases to 89.9% while the average number of false positive hits only increases to 0.9 per random genome. As shown in Figure 3B, typically, a tss cutoff between 35 and 45 can reasonably balance the sensitivity and selectivity. Though the number of false positives is influenced by many factors like GC content of the genome, in our experiments, it is mainly related to the size of the training set and the number of candidate sequences (which is principally determined by the BLAST E-value in the preprocessing step) in the test set. Here, the BLAST E-value 10 was selected in the preprocessing step and the training set contains
1000 miRNAs. Based on this estimation, one can choose a proper tss cutoff for his own application.
Comparison with other methods
We compared miRAlign with BLAST search and ERPIN search (Legendre et al., 2004) on several genomic datasets. All the known pre-miRNA sequences in training sets Train_Sub_1 and Train_Sub_2 were used as BLAST search queries against the C.briggsae genome, with a non-stringent BLAST parameter setting (word length = 7, E-value cutoff = 0.01). Only the BLAST hits that can be fold with MFE lower than –20 kcal/mol were treated as putative miRNAs. Table 1 shows the summary of the results of BLAST search and miRAlign search. With Train_Sub_1, 66 and 71 known miRNAs genes were detected by BLAST search and miRAlign, respectively, 63 of which are identical. The average numbers of false positives are 7.1 and 0.9 using the two methods, respectively. With Train_Sub_2, known miRNAs detected by BLAST search and by miRAlign sharply decreased to 5 and 8, respectively. And the average numbers of false positives per C.briggsae genome slightly reduced to 5.9 and 0.8. In this case, all the five BLAST-hits were also detected by miRAlign. This result illustrates that miRAlign outperforms BLAST search in both sensitivity and selectivity, and furthermore, nearly all the known miRNAs hit by BLAST can be detected by miRAlign also.
The latest version of ERPIN 4.2.5 and the 18 animal miRNA training sets (Legendre et al., 2004) were downloaded from http://tagc.univ-mrs.fr/erpin/, and the recommended parameters for each training set were also got from the ERPIN website. These datasets contain 180 pre-miRNAs from C.briggsae (15), C.elegans (23), D.melanogaster (31), H.sapiens (54) and M.musculus (57). For the sake of comparison, we performed ERPIN and miRAlign search using these 180 miRNAs as training data on five genomes (C.elegans, C.briggsae, D.melanogaster, D.pseudoobscura and G.gallus) and used ERPIN E-value 0.01 and miRAlign tss 35 as our threshold for the putative miRNAs. Table 2 shows the statistics of miRAlign and ERPIN results. Despite that the training set is favorable for ERPIN search (all these miRNAs in the training set can be grouped in large families to construct the profile), miRAlign achieved comparable specificity and higher sensitivity than ERPIN search on all five genomes used in this study, and identified all the known miRNAs searched by ERPIN. Though on these large miRNA families the improvement on sensitivity is not very dramatic, in practical applications, greater improvement can be expected because miRAlign can be applied on all miRNA families but ERPIN only works for larger ones.
|
Prediction of miRNAs in A.gambiae
A.gambiae, which has diverged from D.melanogaster
250 million years ago, is a highly adapted, successful dipterans species(Gaunt and Miles, 2002; Yeates and Wiegmann, 1999). This mosquito is a major vehicle for the transmission of malaria. Up to now, no experimentally identified or computationally predicted A.gambiae miRNAs are reported. We applied our approach on the A.gambiae data using Train_all as the training set. There are totally 1100 non-overlapped candidates getting non-zero tss, of which 59 putative miRNA sequences are detected with tss >35. During our revision of this paper, 38 A.gambiae miRNAs were reported in the MicroRNA Registry release 6.0, and 37 of them were covered by our predictions. This result validates the sensitivity and accuracy of our method. Supplementary Table S2 lists the putative novel miRNAs predictedby miRAlign. |
CONCLUSION |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
We developed a new computational approach called miRAlign to predict new miRNA genes that are homologs or orthologs to known miRNAs. Combining sequence and structure alignments, miRAlign has better performance than previously reported homologue search methods. With the help of this method and other high-throughput computational approaches, systematic genome-wide annotation will gradually reveal the panorama of the miRNAs and help us to further understand the functions and evolutionary footprint of miRNAs.
It should be noted that our experiments in this work are mainly on animal data. We also investigated the feasibility of the application of miRAlign for plant miRNA homologue search. A sensitivity of 70% was achieved by detecting 28 of the 40 known Zea mays miRNAs genes (Maher et al., 2004) with A.thaliana and O.sativa miRNAs as the training set (data not shown). This result indicates that miRAlign can be also applied in plants and further investigation is undergoing.
|
Acknowledgments |
|---|
We appreciate Matthias Hochsmann for providing the source code for RNAforester. Thanks to Ying Huang, Jun Cai and Chenghai Xue for helpful discussions. We also thank Prof. Yunshen Sun for checking the language. This project is supported in part by NSFC (grants 60171038, 60234020, 60405001) and the National Basic Research Program (2004CB518605) of China.
Conflict of Interest: none declared.
|
Footnotes |
|---|
The authors wish it to be known that, in their opinion, the first two authors should be regarded as joint First Authors. 
Received on March 23, 2005; revised on June 25, 2005; accepted on June 27, 2005
|
REFERENCES |
|---|
|
TOPAbstractINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONREFERENCES |
|---|
Altschul, S.F., et al. (1990) Basic local alignment search tool. J. Mol. Biol., 215, 403–410[CrossRef][ISI][Medline].
Ambros, V., et al. (2003) A uniform system for microRNA annotation. RNA, 9, 277–279
Bartel, B. and Bartel, D.P. (2003) MicroRNAs: at the root of plant development? Plant Physiol., 132, 709–717
Bartel, D.P. (2004) MicroRNAs: genomics, biogenesis, mechanism, and function. Cell, 116, 281–297[CrossRef][ISI][Medline].
Bonnet, E., et al. (2004) Detection of 91 potential conserved plant microRNAs in Arabidopsis thaliana and Oryza sativa identifies important target genes. Proc. NatlAcad. Sci. USA, 101, 11511–11516
Gaunt, M.W. and Miles, M.A. (2002) An insect molecular clock dates the origin of the insects and accords with palaeontological and biogeographic landmarks. Mol. Biol. Evol., 19, 748–761
Gautheret, D. and Lambert, A. (2001) Direct RNA motif definition and identification from multiple sequence alignments using secondary structure profiles. J. Mol. Biol., 313, 1003–1011[CrossRef][ISI][Medline].
Griffiths-Jones, S. (2004) The MicroRNA Registry. Nucleic Acids Res., 32, D109–D111
Grishok, A., et al. (2001) Genes and mechanisms related to RNA interference regulate expression of the small temporal RNAs that control C.elegans developmental timing. Cell, 106, 23–34[CrossRef][ISI][Medline].
Hammond, S.M., et al. (2000) An RNA-directed nuclease mediates post-transcriptional gene silencing in Drosophila cells. Nature, 404, 293–296[CrossRef][Medline].
He, L. and Hannon, G.J. (2004) MicroRNAs: small RNAs with a big role in gene regulation. Nat. Rev. Genet., 5, 522–531[CrossRef][Medline].
Hochsmann, M., Toller, T., Giegerich, R., Kurtz, S. (2003) Local similarity in RNA secondary structures. Proceedings of the IEEE Computational Systems Bioinformatics Conference , Stanford, CA , pp. 159–168.
Hofacker, I., et al. (1994) Fast folding and comparison of RNA secondary structures. Monatshefte f. Chemie, 125, 167–188[CrossRef].
Hutvagner, G., et al. (2001) A cellular function for the RNA-interference enzyme Dicer in the maturation of the let-7 small temporal RNA. Science, 293, 834–838
Jones-Rhoades, M.W. and Bartel, D.P. (2004) Computational identification of plant microRNAs and their targets, including a stress-induced miRNA. Mol. Cell, 14, 787–799[CrossRef][ISI][Medline].
Ketting, R.F., et al. (2001) Dicer functions in RNA interference and in synthesis of small RNA involved in developmental timing in C. elegans. Genes Dev., 15, 2654–2659
Klein, R.J. and Eddy, S.R. (2003) RSEARCH: finding homologs of single structured RNA sequences. BMC Bioinformatics, 4, 44[CrossRef][Medline].
Lagos-Quintana, M., et al. (2001) Identification of novel genes coding for small expressed RNAs. Science, 294, 853–858
Lai, E.C. (2003) MicroRNAs: runts of the genome assert themselves. Curr Biol., 13, R925–R936[CrossRef][ISI][Medline].
Lai, E.C., et al. (2003) Computational identification of Drosophila microRNA genes. Genome Biol., 4, R42[CrossRef][Medline].
Lau, N.C., et al. (2001) An abundant class of tiny RNAs with probable regulatory roles in Caenorhabditis elegans. Science, 294, 858–862
Lee, R.C. and Ambros, V. (2001) An extensive class of small RNAs in Caenorhabditis elegans. Science, 294, 862–864
Lee, R.C., et al. (1993) The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell, 75, 843–854[CrossRef][ISI][Medline].
Lee, Y., et al. (2002) MicroRNA maturation: stepwise processing and subcellular localization. EMBO J., 21, 4663–4670[CrossRef][ISI][Medline].
Lee, Y., et al. (2003) The nuclear RNase III Drosha initiates microRNA processing. Nature, 425, 415–419[CrossRef][Medline].
Legendre, M., et al. (2004) Profile-based detection of microRNA precursors in animal genomes. Bioinformatics, 21, 841–845[Medline].
Lim, L.P., et al. (2003a) Vertebrate microRNA genes. Science, 299, 1540
Lim, L.P., et al. (2003b) The microRNAs of Caenorhabditis elegans. Genes Dev., 17, 991–1008
Maher, C., Timmermans, M., Stein, L., Ware, D. (2004) Identifying microRNAs in plant genomes. Proceedings of the 2004 IEEE Computational Systems Bioinformatics Conference (CSB 2004) , Stanford, CA , pp. 718–723.
Ohler, U., et al. (2004) Patterns of flanking sequence conservation and a characteristic upstream motif for microRNA gene identification. RNA, 10, 1309–1322
Pasquinelli, A.E., et al. (2000) Conservation of the sequence and temporal expression of let-7 heterochronic regulatory RNA. Nature, 408, 86–89[CrossRef][Medline].
Thompson, J.D., et al. (1994) CLUSTAL W: improving the sensitivity of progressive multiple sequence alignment through sequence weighting, position-specific gap penalties and weight matrix choice. Nucleic Acids Res., 22, 4673–4680
Wang, X.J., et al. (2004) Prediction and identification of Arabidopsis thaliana microRNAs and their mRNA targets. Genome Biol., 5, R65[CrossRef][Medline].
Weber, M.J. (2005) New human and mouse microRNA genes found by homology search. FEBS J., 272, 59–73[CrossRef][Medline].
Yeates, D.K. and Wiegmann, B.M. (1999) Congruence and controversy: toward a higher-level phylogeny of Diptera. Annu Rev Entomol, 44, 397–428[CrossRef][ISI][Medline].
Zuker, M. and Stiegler, P. (1981) Optimal computer folding of large RNA sequences using thermodynamics and auxiliary information. Nucleic Acids Res., 9, 133–148
CiteULike 
Connotea 
Del.icio.us What's this?
This article has been cited by other articles:
|
|
 |
|
  X. Wang, J. Gu, M. Q. Zhang, and Y. Li Identification of phylogenetically conserved microRNA cis-regulatory elements across 12 Drosophila species Bioinformatics, January 15, 2008; 24(2): 165 - 171. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. Terai, T. Komori, K. Asai, and T. Kin miRRim: A novel system to find conserved miRNAs with high sensitivity and specificity RNA, December 1, 2007; 13(12): 2081 - 2090. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. Winter, S. Edaye, A. Huttenhofer, and C. Brunel Anopheles gambiae miRNAs as actors of defence reaction against Plasmodium invasion Nucleic Acids Res., November 29, 2007; 35(20): 6953 - 6962. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. Jiang, H. Wu, W. Wang, W. Ma, X. Sun, and Z. Lu MiPred: classification of real and pseudo microRNA precursors using random forest prediction model with combined features Nucleic Acids Res., July 13, 2007; 35(suppl_2): W339 - W344. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. L. S. Ng and S. K. Mishra De novo SVM classification of precursor microRNAs from genomic pseudo hairpins using global and intrinsic folding measures Bioinformatics, June 1, 2007; 23(11): 1321 - 1330. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
W. Ritchie, M. Legendre, and D. Gautheret RNA stem-loops: To be or not to be cleaved by RNAse III RNA, April 1, 2007; 13(4): 457 - 462. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 W. Zhu and C. R. Buell Improvement of whole-genome annotation of cereals through comparative analyses Genome Res., March 1, 2007; 17(3): 299 - 310. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. NG Kwang Loong and S. K. Mishra Unique folding of precursor microRNAs: Quantitative evidence and implications for de novo identification RNA, February 1, 2007; 13(2): 170 - 187. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. Yousef, M. Nebozhyn, H. Shatkay, S. Kanterakis, L. C. Showe, and M. K. Showe Combining multi-species genomic data for microRNA identification using a Naive Bayes classifier Bioinformatics, June 1, 2006; 22(11): 1325 - 1334. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. Dezulian, M. Remmert, J. F. Palatnik, D. Weigel, and D. H. Huson Identification of plant microRNA homologs Bioinformatics, February 1, 2006; 22(3): 359 - 360. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||