ISSN: 0974-276X
Journal of Proteomics & Bioinformatics
Like us on:
Make the best use of Scientific Research and information from our 700+ peer reviewed, Open Access Journals that operates with the help of 50,000+ Editorial Board Members and esteemed reviewers and 1000+ Scientific associations in Medical, Clinical, Pharmaceutical, Engineering, Technology and Management Fields.
 
Meet Inspiring Speakers and Experts at our 3000+ Global Conferenceseries Events with over 600+ Conferences, 1200+ Symposiums and 1200+ Workshops on
Medical, Pharma, Engineering, Science, Technology and Business

Performance and Evaluation of MicroRNA Gene Identification Tools

Swati Sinha1*, T.S. Vasulu1, and Rajat K. De2*
1Biological Anthropology Unit, Indian Statistical Institute, Kolkata, India
2Machine Intelligence Unit, Indian Statistical Institute, Kolkata, India
Corresponding Authors : Dr. Swati Sinha
Biological Anthropology Unit,
Indian Statistical Institute,
Kolkata-108, India,
Tel       : 91-33-25753215,
E-mail : swati.6783@gmail.com
  Dr. Rajat K De
Machine Intelligence Unit,
Indian Statistical Institute,
Kolkata-108, India
Tel       : 91-33-25753105,
Fax      : +91-33-25753026,
E-mail : rajat@isical.ac.in
Received July 02, 2009; Accepted August 11, 2009; Published August 12, 2009
Citation: Sinha S, Vasulu TS, De RK (2009) Performance and Evaluation of MicroRNA Gene Identification Tools. J Proteomics Bioinform 2: 336-343. doi:10.4172/jpb.1000093
Copyright: © 2009 Sinha S, et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Related article at
DownloadPubmed DownloadScholar Google

Visit for more related articles at Journal of Proteomics & Bioinformatics

Keywords
MiPred; Triplet-SVM; BayesMiRNAfind; OneClassmiRNAfind; BayesSVMmiRNAfind; Sensitivity; Specificity; Accuracy; Mathew’s correlation coefficient; Positive predictive value
Abbreviations
miRNA: MicroRNA; pre-miRNA: Precursor MicroRNA; HMM: Hidden Markov Model; SVM: Support
Vector Machine; PCA: Principal Component Analysis; K-NN: K-Nearest Neighbor; MCC: Mathew’s Correlation Coefficient; PPV: Positive Predictive Value
Introduction
Interest in miRNAs and their role as gene expression regulators has been growing immensely (Clop et al., 2006; Feng et al., 2009). The first effort that could identify such a small regulator, the lin-4 RNA in C. elegans, was done by Victor Ambros and colleagues, Rosalind Lee and Rhonda Feinbaum (Bartel, 2004). It was shown that the 21 nt lin-4 RNA, represses mRNA and controls part of the C. elegans larval development. The next small regulatory RNA to be discovered was the let-7, which controls another later developmental stage of C. elegans (Lee et al., 1993; Wightman et al., 1993). They were previously known as small temporal RNAs (stRNAs), but today recognized as the first of the large class of small regulatory non-coding RNA molecules, ‘microRNAs’. Now it is believed that this class of molecules is not only limited to development but also plays a very important role in the regulation of a wide range of biological processes (Gard et al., 2006; Feng et al., 2009).
MicroRNAs are small non-coding RNAs of approximately 22nt (ranged 19-25nt) known to be involved in posttranscriptional gene regulation either by cleavage of mRNAs or repressing the translation of mRNAs (Bartel, 2004). The microRNAs reported to be encoded within noncoding regions of genomes and within protein coding genes. The miRNA genes are transcribed by RNA polymerase II; in some cases RNA polymerase III may also be involved. Primary transcripts of the miRNA genes, ‘primiRNAs’, are processed in the nucleus to ‘pre-miRNAs’ by the RNAase III type endonuclease ‘Drosha’ and exported from the nucleus to the cytoplasm by nuclear export factor Exportin 5 and the Ran-GTP cofactor. The 60– 90-nt miRNA precursors form the stem-loop structures, and the cytoplasmic ribonuclease class III enzyme ‘Dicer’ excises miRNAs from the pre-miRNA hairpin stem (Bartel, 2004). Dicer, either alone or with the help of Drosha, cleaves both strands of the precursor to form a doublestranded miRNA/miRNA* duplex (Dezulian et al., 2006). The mature miRNA strand of the duplex with the less stable 5’ is then incorporated into RNA induced silencing complex (RISC) while the other strand is rapidly degraded.
Since the early studies elucidating the important role of miRNAs, there has been a continuous increase in the number of microRNAs along with simultaneous increase in the range of genomes encoding miRNAs, and it is worthy to mention that these small regulators play much more important role than previously thought (Gard et al., 2006). MicroRNAs and their associated proteins appear to be one of the more abundant ribo-nucleoprotein complexes in the cell. In general these have been detected by molecular genetic technique of cloning and Northern blotting but the detection of miRNAs whose expression is restricted to non-abundant cell types or specific environmental conditions could still be missed. Moreover these experimental methods are biased towards abundant microRNAs and are time-consuming. In these regard the computational approaches have been developed to complement experimental approaches to miRNA gene identification. It was found that the prediction of miRNA precursor candidates is fairly easy, whereas excluding false positives, as well as, exact prediction of the mature miRNA is a harder task. Thus, the computational detection methods must therefore be refined to serve as a faster, better, and more efficient method for miRNA detection.
In this regard, so far, several algorithms related to miRNA gene identification have been developed successfully. These approaches possess different roles like to predict the miRNA genes based on the evolutionary conservation of miRNA (miRScan, MirSeeker, MirFinder etc). Similarly various other bioinformatics predictions combined with microarray analysis (PalGrade); HMM (ProMir) and structure-and-sequence analysis (miRAlign, MicroHarvester) helped to develop these different algorithms. Concept of machine learning approaches also was incorporated with the biology of miRNA to develop algorithms for the same purpose of miRNA gene identification, as for example: Diana-MicroH, mirCoS, miMatcher, Microprocessor-SVM, miRNA-SVM, RNAmicro, miREncoding etc. Apart from these algorithms, some other approaches do make predictions at different step of miRNA biogenesis e.g., some deal with the problem of classification, especially whether a given RNA sequence is miRNA precursor or not (‘Triplet-SVM’ and ‘MiPred’); and some other approaches predict the location of mature miRNA in a given gene sequence (‘BayesMiRNAfind’, ‘OneClassmiRNAfind’, ‘BayesSVMmiRNAfind’ etc). Based on these two criteria, we have attempted to investigate the performance of the various tools of miRNA gene identification in this study. The comparative analysis can tell how well we can predict a given RNA sequence to be a precursor-miRNA and then how well we may know what is the location of mature miRNA in that particular precursor sequence.
Methodologies for Gene Identification
For the computational identification of miRNA gene various successful methods had been developed so far and it is a major thirst research area. This is because these tiny gene regulators have so many important functions such as their role in developmental timing, cell death, cell proliferation, haematopoesis and patterning of nervous system (Ambros, 2004). Lim et al., (2003) came up with an idea to identify miRNA genes that are conserved in more than one genome and developed miRScan for the same purpose. It was followed by another algorithm MirSeeker that analyzes the intronic and intergenic regions that are evolutionary conserved in D. melanogaster and D. pseudoobscura (Lai et al., 2003). The conservedness of these short sequences across species seems to be an important factor to develop an algorithm for miRNA gene prediction. Accordingly, Bonnet et al., (2004) developed MirFinder, a genome wide computational approach, to detect miRNA genes present in the Arabidopsis genome based on the conservation of sequences between Arabidopsis and Oryza genome. One more approach for the identification of miRNAs in A. thaliana was based on rigid complementarities that exist between plant miRNAs and their targets named as findMiRNA (Adai et al., 2005).
Apart from sequence conservation, other factors also helped to develop different algorithms for the same purpose. In 2005, Bentwitch developed PalGrade, which is an integrative approach that combines bioinformatics prediction with microarray analysis and sequence directed cloning. Information both at sequence and structural level helped to develop ProMir which is a probabilistic co-learning method based on paired HMM to identify close as well as distant homolog (Nam et al., 2005). Wang et al., (2005) were able to detect new miRNA based on structure and sequence alignment with a novel computational approach miRAlign. Similar sequence and structure information helped to develop MicroHarvester for the identification of candidate miRNA homolog in a set of sequences given a query miRNA (Dezulian et al., 2006). Another concept, that miRNAs are often found in clusters, was the basis of mirAbela (Sewer et al., 2005) that took into consideration only those genomic regions that are present around known miRNAs from mouse, human and rat.
Novel concepts kept on floating and led to novel discoveries, Berezikov et al., (2005) reported that phylogenetic shadowing of miRNAs in primate species revealed a characteristic conservation profile that can be used to detect the majority of known miRNAs efficiently and also predict an extensive novel set of miRNAs based on human- mouse-rat genome wide comparisons. On the other hand Lindow et al., (2005), developed ‘microMatcher’ for the identification of plant miRNAs that do not depend on phylogenetic conservation and identified 592 novel miRNAs which were not conserved in other plant genomes.
This is the present scenario where several prominent computerized miRNA detection approaches have been developed and utilized successfully. Most of these predictor algorithms depend on evolutionary conservation of miRNA sequences between different species. Such approaches allow filtering out many of the false-positive candidates, but they are obviously limited to detecting only the conserved microRNAs. Hence the concept of machine learning came into the picture when the need was to identify the non-conserved miRNAs. Such way of identification of novel microRNAs is a difficult pattern-recognition challenge. A single property is not sufficient for accurately detecting microRNAs, and in most cases rigid thresholds of the values for each of those properties is also not sufficiently sensitive. Rather, it is the combination of multiple properties, along with suitably different weighing of these different properties, that provides desirable accuracy. In this regard, many prediction algorithms based on machine learning have been developed. Xue et al., (2005), first developed an ab initio method called as Triplet-SVM, for distinguishing true pre-miRNAs from other pre-miRNA like hairpin structures taking into account a novel local contiguous structure sequence feature and used SVM with these features to classify real and pseudo pre-miRNAs. In order to improve the predictions by Triplet-SVM, another tool MiPred was developed which utilized the same feature (local contiguous structure sequence feature) in a hybrid way with another features including the MFE of the secondary structure, dinucleotide shuffling and P-Value of randomization test with a novel machine learning algorithm: Random Forest (Jiang et al., 2007). Several other algorithms made use of the SVM like Diana-MicroH (Szafranski et al., 2006) to predict miRNA hairpins using unique feature related to enzymatic cleavage with two additional features viz., GC content and stem linearity; mirCoS (Sheng et al., 2007) that used three SVM models sequentially to discover novel miRNAs in mammalian genomes; Microprocessor-SVM that predicts processing sites for 50% of known human 5’ miRNAs and miRNA-SVM that is trained on the output of the former one to identify non-conserved miRNAs (Helvik et al., 2007); RNAmicro, another SVM based approach that includes non stringent filter for consensus secondary structures and can easily identify pre-miRNAs in multiple sequence alignment (Hertel et al., 2006); miMatcher pipeline developed by Lindow et al.,(2007) performs intragenomic matching of potential miRNAs and their targets followed by classification of these miRNA candidates using SVM and miREncoding (Zheng et al., 2006) to encode the pre-miRNA sequences together with their secondary structures into the proposed 43 features using Weka software (Frank et al., 2004) in order to evaluate the performance of the selected classification algorithms along with the use of polynomial kernels for SVM algorithm. Apart from SVM Bayes classifier had also been used to develop new machine learning algorithm BayesMiRNAfind (Yousef et al., 2006), BayesSVMmiRNAfind (http://wotan.wistar.upenn.edu/ BayesSVMmiRNAfind/) and OneClassMiRNAfind gene algorithm (Yousef et al., 2008). Brameier et al., (2007) used, for the first time, machine learning algorithm based on linear genetic programming and developed a unique ab initio method called as mirPRed for the prediction of novel mature miRNAs by genome scanning.
Materials and Methodology
Materials
The comparative study used five tools falling into two different categories. The first one includes ‘MiPred’ and ‘Triplet-SVM’ meant for the classification of real and pseudo pre-miRNAs and the second one includes ‘BayesMiRNAfind’, ‘OneClassmiRNAfind’ and ‘BayesSVMmiRNAfind’ for the prediction of mature miRNA. Out of the three tools in the latter category ‘OneClassMiRNAfind’ have options for five classifiers viz. SVM, Gaussian, Kmeans, PCA and K-NN and ‘BayesSVMmiRNAfind’ have options for two classifiers viz. naïve-Bayes and SVM. Positive as well as negative datasets were used for this comparative study. The positive dataset consists of 678 real miRNA precursor sequences of Homo sapiens, which were downloaded from the miRBase database release 11.0 (Griffiths-Jones et al., 2008, Griffiths-Jones et al., 2006, Griffiths-Jones, 2004). These sequences in the database have been either experimentally supported or obtained from literature mining and thus are the actual pre-miRNAs. The use of this positive dataset will therefore help to identify the number of true positives and false negatives that will define the sensitivity of a particular tool.
The negative dataset consists of 700 sequences of human pseudo pre-miRNAs. The source of this negative dataset is the coding dataset used by Xue et al., (2005). This ‘coding dataset’ consists of 8494 pre-miRNA like hairpins from which we collected the first 700 sequences (in order to make the number of positive and negative dataset sequences approximately equal) for this comparative analysis of various tools. These 700 sequences are the pre-miRNA like hairpins which are basically those sequence segments that have similar stem loop structure as actual miRNAs but still not been reported as premiRNAs (Xue et al., 2005). The use of this negative dataset will help to identify the number of false positives and true negatives that will define the specificity of a particular tool. Further details regarding the formation of the coding dataset can be obtained from Xue et al., (2005). The number of sequences given, as input, to all of these tools is same except for Triplet-SVM, where some sequences are not accepted due to the constraints in the algorithm.
Analysis
The positive and negative datasets are given as input to each of these tools and the output was analyzed to calculate the number of true positives (TP), true negatives (TN), false positives (FP) and false negatives (FN).
True positives is defined as the number of experimentally supported miRNA precursors that are predicted by a program and false negatives are those experimentally supported miRNA precursors that are not predicted by the program. Similar to the above case, false positives is defined as the number of all negatives that is predicted by a program and true negative on the other hand is all negatives that is not predicted by the program (Sethupathy et al., 2006).
Further, in order to evaluate the performance of these different predictive tools, we used the statistical parameters, viz., Sensitivity (Se), Specificity (Sp), Accuracy (Acc),a summary statistic: Mathew correlation coefficient (MCC) and Positive predictive value (PPV). These parameters are based on TP, FN, TN and FP and are calculated using the following equations (Jiang et al., 2007):
Accuracy (Acc) = (TP+TN)/(TP+TN+FP+FN)*100
Specificity (Sp) = TN/(TN+FP)*100
Sensitivity (Se) = TP/(TP+FN)*100
MCC=((TP*TN)-(FP*FN))/
((TP+FP)*(TN+FN)*(TP+FN)*(TN+FP))½
The Positive Predictive value (PPV) is calculated using the following formula (http://www.medicine.uiowa.edu/path_handbook/Appendix/Chem/PRED_VALUE_THEORY.html).
Positive Predictive value (PPV) = TP/ (TP+FP) * 100.
Results
The results of the miRNA prediction tools are summarized in two ways; first in terms of percent count of TP, FN, FP and TN (Table 1) and second in terms of Sensitivity, Specificity, Accuracy, MCC and PPV of each tool (Table 2).
Table 1 shows the percent count of TP, FN, FP and TN obtained by the two types of predictive algorithms; a) Prediction of real/pseudo miRNA precursor and b) Prediction of mature miRNAs based on both positive and negative data sets.
Table 2 shows the performance evaluation indicators, especially Sensitivity, Specificity, Accuracy, MCC and PPV obtained by the same two types of predictive algorithms: a) Prediction of real/pseudo miRNA precursor and b) Prediction of mature miRNA.
Prediction of Real/pseudo miRNA Precursor
Of the two methods considered for classification of real and pseudo pre-miRNAs, MiPred is more sensitive in identifying the pseudo precursor miRNAs (96%), whereas TripletSVM is less accurate in identifying both types of miRNAs (Table 1a). On the other hand, the Specificity of MiPred is as high as 96% compared to the 85% specificity of Triplet-SVM. The results are in agreement to the total number of true negatives given by the respective tools. The Sensitivity of both these tools is almost same (78- 80%) but because of the high specificity of MiPred the overall accuracy is more for this tool. Hence MiPred is a better tool when compared to Triplet-SVM which is also indicted by the 95.12% PPV for this tool. The MCC value for MiPred (0.7761) is high compared to the other tool TripletSVM (0.6362) showing the high efficiency of MiPred in classifying the real and pseudo pre-miRNA sequences (Table 2a).
Prediction of Mature miRNA
In case of predictive algorithms related to prediction of mature miRNAs, the two methods OneClassmiRNAfind and BayesMiRNAfind show similar results of correctly identifying the pseudo miRNA data (86-96%) than the real miRNA data (69-80%). Except the SVM classifier of OneClassmiRNAfind all other classifiers show relatively high percent of FP (13.43%) whereas the BayesMiRNAfind shows a lower FP value (10.43%), which is higher than that of one-class SVM classifier, but lower than all other one-class classifiers. Interestingly, the tool BayesSVMmiRNAfind was able to identify the real pre-miRNAs (87-96%) but is least efficient in identifying the pseudo pre-miRNA data with a TN value of 8 % only (Table 1b). On the other hand, the specificity of one class SVM classifier is best (96%) followed by BayesMiRNAfind (89.57%), whereas the specificities of all other one-class classifiers are in the same range (86.57%). Again one important observation is the very low specificity of BayesSVMmiRNAfind (8%) suggesting that this tool is not the preferred choice to correctly identify especially the pseudo pre-miRNAs. The sensitivities of the three tools along with their classifiers fall in the range of 71-87% except for BayesSVMmiRNAfind where it is 96%. This suggests that although BayesSVMmiRNA find is far from identifying the negative data correctly, but it can very well be used to predict the positive data. Further, the positive predictive value (PPV) of this tool is also low for both of its classifiers (for SVM it is 47.94% and for naïve Bayes it is 50.46%) when compared to other tools for the prediction of mature miRNA depicting the ineffectiveness of the predictions made by it. The overall accuracy of these tools is somewhat in the same range of 78-83% except for BayesSVMmiRNAfind, where there is a decrease in the range 47-51% because of its very low specificity. Further, the MCC value, which tells about the efficiency of the tool, falls in the same range (0.571-0.6768) but it is lowest in case of BayesSVMmiRNAfind for both of its classifiers naïve-Bayes (0.0853) and SVM (0.1029) (Table 2b).
Discussion
The basic principle of many computational methods is to learn from known examples in order to find new ones and make better predictions. From a computational perspective, this corresponds to the problem of machine learning, an area of artificial intelligence used to develop techniques that allow computers to learn from examples. Since all the mechanisms behind miRNAs and their actions are not completely revealed, computational tasks associated with miRNA studies are often posed as a challenging machine learning problem with limited prior information (Yoon et al., 2006). In spite of such difficulties several algorithms based on the concept of machine learning have been developed but because the field of miRNA research is still in its blooming phase and the understanding at the molecular level is yet not very clear, the process of algorithms development for miRNA identification may not be completely exhaustive. It demands more understanding of the molecular aspect of miRNA biology and more clarity is needed to develop more accurate and efficient tools for the aforesaid purpose. Thus a comparative analysis of the performance of the various prediction tools available might be useful to carry out further research work in this area.
The present study takes into consideration the comparative analysis of the tools available for pre-miRNA classification and mature miRNA prediction in order to understand the limitations in these algorithms so that further efforts can be done for their improvement. Of all the tools studied, the earliest was TripletSVM, which included just the local contiguous sequence-structure feature. There is no inclusion of any thermodynamic related features hence there is a possibility of improvement in its overall performance. And it was achieved when MiPred was developed later on, which includes a hybrid feature incorporating local contiguous sequence structure feature along with the MFE and P value of randomization test. This improvement is visible in the result (Table 2) as the sensitivity as well as accuracy of MiPred is high compared to that of TripletSVM. Moreover, the specificity of MiPred is ~11% higher than TripletSVM; this is of much importance, as specificity is related to the false positive rate. More the specificity more will be the tendency of the program not to erroneously predict the negative data. The high performance of MiPred can also be contributed to the novel classifier algorithm Random Forest used in their program along with SVM.
Among the rest three tools, BayesMiRNAfind shows good specificity and accuracy indicating its better performance. The reason behind might be the inclusion of certain rules based on miRNA gene structure and sequence, thereby allowing prediction of non-conserved miRNAs. Moreover in order to reduce the false positive rate the tool is based on a comparative analysis over multiple species in an attempt to develop an algorithm that has higher specificity but almost similar sensitivity, which is also depicted by the results. OneClassMiRNAfind with SVM shows the best specificity among all classifiers which themselves have the same specificities. In case of sensitivities, OC-KNN and OC-K-means are superior to others as measured by their ability to capture only the known miRNAs. Two-class classifier approach BayesSVMmiRNAfind with SVM is showing the best sensitivity among all the tools compared but it has a very low specificity due to which the overall predictive criterion of accuracy is not good. Thus, BayesSVMmiRNAfind with SVM as a classifier has the highest sensitivity but its specificity is lowest (Figure 1).
Keeping in mind the various algorithms and methodologies developed so far one possible area of further research is to incorporate certain new features related to miRNA and to develop some new and more efficient algorithm for the same purpose. Furthermore, one possible limitation of the present study is that it is only based on comparing single programs and we have not considered the possibilities of combinations of several programs e.g., performance of various unions and intersections of individual programs, which might lead to a better comparative analysis.
Conclusions
There have been several computational methods developed to predict miRNAs. Five different methods based on two different categories of miRNA gene identification tools have been compared to understand their relative performance. Among all the tools, MiPred shows the best performance. One class approach can be a good alternative but as far as overall accuracy is concerned, certain improvements need to be incorporated for a better performance. Moreover, BayesSVMmiRNAfind using SVM and naïve Bayes classifier show lowest specificity although the sensitivity is quite high in both the cases.
References
  1. Adai A, Johnson C, Mlotshwa S, Evans SA, Manocha V, et al. (2005) Computational prediction of miRNAs in Arabidopsis thaliana. Genome Research, 15: 78-91. » CrossRef   » PubMed  »  Google Scholar
  2. Bartel DP (2004) MicroRNAs: genomics, biogenesis, mechanism, and function. Cell 116: 281-297. » CrossRef   » PubMed  »  Google Scholar
  3. Bentwich I (2005) Prediction and validation of microRNAs and their targets. FEBS Lett 579: 5904 - 5910. » CrossRef   » PubMed  »  Google Scholar
  4. Berezikov E, Guryev V, Belt JV, Weinholds E, Plasterk RHA, et al. (2005) Phylogenetic Shadowing and Computational Identification Of Human microRNA genes. Cell 120: 21-24. » CrossRef   » PubMed  »  Google Scholar
  5. Bonnet E, Wuyts J, Rouze P, Peer YV (2004) Detection of 91 potential conserved plant microRNAs inArabidopsis thaliana and Oryza sativa identify important target genes. PNAS 101: 11511-11516. » CrossRef  »  Google Scholar
  6. Brameier M and Wiuf C (2007) Ab initio identification of human microRNAs based on structure motifs. BMC Bioinformatics 8: 478. » CrossRef   » PubMed  »  Google Scholar
  7. Clop A, Marcq F, Takeda H, Pirottin D, Tordoir X (2006) A mutation creating a potential illegitimate microRNA target site in the myostatin gene affects muscularity in sheep. Nature Genetics 38: 813-818. » CrossRef   » PubMed  »  Google Scholar
  8. Dezulian T, Remmert M, Palatnik JF, Weigel D, Huson DH (2006) Identification of plant microRNA homologs. Bioinformatics 22: 359-360. » CrossRef   » PubMed  »  Google Scholar
  9. Dror G, Sorek R, Shamir R (2005) Accurate identification of alternatively spliced exons using support vector machine. Bioinformatics 21: 897-901. » CrossRef   » PubMed  »  Google Scholar
  10. Feng J, Sun G, Yan J, Noltner K, Li W, et al. (2009) Evidence for X-chromosomal schizophrenia assoaciated with miRNA alterations. PLoS ONE 4: e6121. » CrossRef   » PubMed  »  Google Scholar
  11. Frank E, Hall M, Trigg L, Holmes G, Witten I (2004) Data mining in bioinformatics using Weka. Bioinformatics 2479-2481. » CrossRef   » PubMed  »  Google Scholar
  12. Gard OS, Thomassen, Rosok O, Rognes T (2006) Computational Prediction of MicroRNAs Encoded in Viral and Other Genomes. Journal of Biomedicine and Biotechnology Article ID 95270: 1-10. » CrossRef   » PubMed  »  Google Scholar
  13. Griffiths-Jones S, Saini HK, Van DS, Enright AJ (2008) miRBase: tools for microRNA genomics. Nucleic Acid Research 36: D154-D158. » CrossRef   » PubMed  »  Google Scholar
  14. Griffiths-Jones S, Grocock RJ, Van DS, Bateman A, Enright AJ (2006) miRBase: microRNA sequences, targets and gene nomenclature. Nucleic Acid Research 34: D140-D144. » CrossRef   » PubMed  »  Google Scholar
  15. Griffiths-Jones S (2004) The microRNA Registry. Nucleic Acid Research 32: D109-D111. » CrossRef   » PubMed  »  Google Scholar
  16. Gruber AR, Neubock R, Hofacker IL, Washietl S (2007) The RNAz web server: prediction of thermodynamically stable and evolutionarily conserved RNA structures. Nucleic Acids Research 35: W335-W338. » CrossRef   » PubMed  »  Google Scholar
  17. Helvik SA, Snove O, Saetrom P (2007) Reliable prediction of Drosha processing sites improves microRNA gene prediction. Bioinformatics Genome Analysis 23: 142-149. » CrossRef   » PubMed  »  Google Scholar
  18. Hertel JA and Stadler PF (2006) Hairpins in a Haystack: recognizing microRNA precursors in comparative genomics data. Bioinformatics 22: e197-e202. » CrossRef   » PubMed  »  Google Scholar
  19. Hofacker IL (2003) Vienna RNA secondary structure server. Nucleic Acids Research 31: 3429-3431. » CrossRef   » PubMed  »  Google Scholar
  20. Jiang P, Wu H, Wang W, Wa M, Sun X, et al. (2007) MiPred: classification of real and pseudo microRNA precursor using random forest prediction model with combined features. Nucleic acid research, 35: W339- W344. » CrossRef »  Google Scholar
  21. Jones SG (2004) The microRNA Registry. Nucleic Acids Research 32: D109-D111. » CrossRef   » PubMed  »  Google Scholar
  22. Lai EC, Tomancak P, Williams RW, Rubin GM (2003) Computational identification of Drosophila microRNA genes. Genome Biology 4: R42. » CrossRef   » PubMed  »  Google Scholar
  23. Lee RC, Feinbaum RL, Ambros V (1993) The C. elegans heterochronic gene lin-4 encodes small RNAs with antisense complementarity to lin-14. Cell 75: 843- 854. » CrossRef   » PubMed  »  Google Scholar
  24. Lim LP, Lau NC, Weinstein EG, Abdelhakim A, Yekta S, et al. (2003) The microRNAs of Caenorhabditis elegans. Genes & Development 17: 991-1008. » CrossRef   » PubMed  »  Google Scholar
  25. Lindow M and Gorodkin J (2007a) Principles and Limitations Computational MicroRNA Gene and Target Finding. Dna and Cell Biology 26: 339-351. » CrossRef   » PubMed  »  Google Scholar
  26. Lindow M and Krogh A (2005) Computational evidence for hundreds of non-conserved plant microRNAs. BMC Genomics 6: 119. » CrossRef   » PubMed  »  Google Scholar
  27. Lindow M, Jacobsen A, Nygaard S, Mang Y, Krogh A (2007) Intragenomic Matching Reveals a Huge Potential for miRNA-MediatedRegulation in Plants. PLoS Comput Biol 3: 2379-2390. » CrossRef   » PubMed  »  Google Scholar
  28. Nam JW, Shin KR, Han J, Lee Y, Kim VN, et al. (2005) Human microRNA prediction through probabilistic colearning of sequence and structure. Nucleic Acids Research 33: 3570-3581. » CrossRef   » PubMed  »  Google Scholar
  29. Paul SA, Landgraf P, Aravin A, Pfeffer S, Brownstein MJ, et al. (2005) Identification of clustered microRNAs using an ab initio prediction Method. BMC Bioinformatics 6: 267. » CrossRef   » PubMed  »  Google Scholar
  30. Sethupathy P, Megraw M, Hatzigeorgiou AG (2006) A guide through present computational approaches for the identification of mammalian microRNA targets. Nature Methods 3: 881-886. » CrossRef   » PubMed  »  Google Scholar
  31. Sheng Y, Engstrom PG, Lenhard B (2007) Mammalian MicroRNA Prediction through a Support Vector Machine Model of Sequence and Structure. PloS ONE 9: e946. » CrossRef   » PubMed  »  Google Scholar
  32. Szafranski K, Megraw M, Reczko M, Hatzigeorgiou AG Support Vector Machines for predicting microRNA Hairpins. » CrossRef  »  Google Scholar
  33. Wang X, Zhang J, Li F, Gu J, He T, et al. (2005) MicroRNA identification based on sequence and structure alignment. Bioinformatics 21: 3610-3614. » CrossRef   » PubMed  »  Google Scholar
  34. Wightman B, Ha I, Ruvkun G (1993) Posttranscriptional regulation of the heterochronic gene lin-14 by lin-4 mediates temporal pattern formation in C. elegans. Cell 75: 855-862. » CrossRef   » PubMed  »  Google Scholar
  35. Xue C, Li F, He T, Liu GP, Li Y, et al. (2005) Classification of real and pseudo microRNA precursors using local structure-sequence features and support vector machine. BMC Bioinformatics 6: 310. » CrossRef   » PubMed  »  Google Scholar
  36. Yoon S and Micheli GD (2006) Computational Identification of MicroRNAs and Their Targets. Birth Defects Research 78: 118-128. » CrossRef   » PubMed  »  Google Scholar
  37. Yousef M, Jung S, Showe LC, Showe MK (2008) Learning from positive examples when the negative class is undetermined-microRNA gene identification. Algorithms for Molecular Biology 3: 2. » CrossRef   » PubMed  »  Google Scholar
  38. Yousef M, Nebozhyn M, Shatkay H, Kanterakis S, Showe LC, et al. (2006) Combining multi-species genomic data for microRNA identification using a naive Bayes classifier. Bioinformatics 22: 1325-1334. » CrossRef   » PubMed  »  Google Scholar
  39. Zheng Y, Hsu W, Lee ML, Wong LS Exploring Essential Attributes For Detecting MicroRNA Precursors From Background Sequences.
Select your language of interest to view the total content in your interested language
Post your comment

Share This Article

Relevant Topics

Recommended Conferences

Article Usage

  • Total views: 11300
  • [From(publication date):
    August-2009 - Sep 28, 2016]
  • Breakdown by view type
  • HTML page views : 7558
  • PDF downloads :3742
 
 

Post your comment

captcha   Reload  Can't read the image? click here to refresh

OMICS International Journals
 
Make the best use of Scientific Research and information from our 700 + peer reviewed, Open Access Journals
 
 
OMICS International Conferences 2016-17
 
Meet Inspiring Speakers and Experts at our 3000+ Global Annual Meetings
 
 

Contact Us

Agri, Food, Aqua and Veterinary Science Journals

Dr. Krish

agrifoodaquavet@omicsinc.com

1-702-714-7001 Extn: 9040

Clinical and Biochemistry Journals

Datta A

clinical_biochem@omicsinc.com

1-702-714-7001Extn: 9037

Business & Management Journals

Ronald

business@omicsinc.com

1-702-714-7001Extn: 9042

Chemical Engineering and Chemistry Journals

Gabriel Shaw

chemicaleng_chemistry@omicsinc.com

1-702-714-7001 Extn: 9040

Earth & Environmental Sciences

Katie Wilson

environmentalsci@omicsinc.com

1-702-714-7001Extn: 9042

Engineering Journals

James Franklin

engineering@omicsinc.com

1-702-714-7001Extn: 9042

General Science and Health care Journals

Andrea Jason

generalsci_healthcare@omicsinc.com

1-702-714-7001Extn: 9043

Genetics and Molecular Biology Journals

Anna Melissa

genetics_molbio@omicsinc.com

1-702-714-7001 Extn: 9006

Immunology & Microbiology Journals

David Gorantl

immuno_microbio@omicsinc.com

1-702-714-7001Extn: 9014

Informatics Journals

Stephanie Skinner

omics@omicsinc.com

1-702-714-7001Extn: 9039

Material Sciences Journals

Rachle Green

materialsci@omicsinc.com

1-702-714-7001Extn: 9039

Mathematics and Physics Journals

Jim Willison

mathematics_physics@omicsinc.com

1-702-714-7001 Extn: 9042

Medical Journals

Nimmi Anna

medical@omicsinc.com

1-702-714-7001 Extn: 9038

Neuroscience & Psychology Journals

Nathan T

neuro_psychology@omicsinc.com

1-702-714-7001Extn: 9041

Pharmaceutical Sciences Journals

John Behannon

pharma@omicsinc.com

1-702-714-7001Extn: 9007

Social & Political Science Journals

Steve Harry

social_politicalsci@omicsinc.com

1-702-714-7001 Extn: 9042

 
© 2008-2016 OMICS International - Open Access Publisher. Best viewed in Mozilla Firefox | Google Chrome | Above IE 7.0 version