Prediction of piRNAs using transposon interaction and a support vector machine
© Wang et al.; licensee BioMed Central. 2014
Received: 2 July 2014
Accepted: 11 December 2014
Published: 30 December 2014
Piwi-interacting RNAs (piRNAs) are a class of small non-coding RNA primarily expressed in germ cells that can silence transposons at the post-transcriptional level. Accurate prediction of piRNAs remains a significant challenge.
We developed a program for piRNA annotation (Piano) using piRNA-transposon interaction information. We downloaded 13,848 Drosophila piRNAs and 261,500 Drosophila transposons. The piRNAs were aligned to transposons with a maximum of three mismatches. Then, piRNA-transposon interactions were predicted by RNAplex. Triplet elements combining structure and sequence information were extracted from piRNA-transposon matching/pairing duplexes. A support vector machine (SVM) was used on these triplet elements to classify real and pseudo piRNAs, achieving 95.3 ± 0.33% accuracy and 96.0 ± 0.5% sensitivity. The SVM classifier can be used to correctly predict human, mouse and rat piRNAs, with overall accuracy of 90.6%. We used Piano to predict piRNAs for the rice stem borer, Chilo suppressalis, an important rice insect pest that causes huge yield loss. As a result, 82,639 piRNAs were predicted in C. suppressalis.
Piano demonstrates excellent piRNA prediction performance by using both structure and sequence features of transposon-piRNAs interactions. Piano is freely available to the academic community at http://ento.njau.edu.cn/Piano.html.
KeywordspiRNAs piRNA prediction Support vector machine (SVM) Chilo suppressalis Drosophila melanogaster Homo sapiens Mus musculus Rattus norvegicus
Non-coding RNAs (ncRNAs) are important RNA molecules. Although they do not encode proteins, their roles in gene regulation are crucial ,. There are many types of long ncRNAs whose functions remain largely unknown . Short ncRNAs, such as microRNAs (miRNAs) and piwi-interacting RNAs (piRNAs), are important post-transcriptional regulators . piRNAs are produced from un-characterized precursors in both male and female germline cells. The discovery of piRNAs was a highly important break-through as they are involved in germ cell formation, germline stem cell maintenance, spermatogenesis and oogenesis -.
The biogenesis of piRNAs is quite different from that of miRNAs. Although details of their biogenesis are currently unclear, several models have been proposed. In germline cells, piRNAs can be produced by the primary processing pathway and by a feed-forward loop, called the “ping-pong” pathway, which uses primary piRNAs to direct cleavage of complementary transposon sense transcripts . These mature sense piRNAs will target complementary antisense piRNA precursors to create mature antisense piRNAs that can continue sense piRNA generation. piRNAs lack apparent structural motif and sequence conservation across different species, making their prediction a difficult task. piRNAs are generally understood to participate in transposon silencing during embryo development . The majority of piRNAs are antisense to transposons. In the genome, piRNAs tend to occur in clusters and to be located in intergenic regions . However, piRNAs are also found in somatic cells , and studying piRNA functionality is still a challenging task because of the wide variation of piRNA sequences.
piRNAs have been reported in human , mouse , rat , zebra fish , and fruit fly . A typical experimental procedure to obtain piRNA data relies on immunoprecipitation of small RNAs bound to the protein PIWI and deep sequencing. However, with this method, it is still hard to identify piRNAs expressed at low levels or with restricted spatiotemporal expression. Therefore, computational prediction can provide an alternative approach to identify potential piRNAs. Unfortunately, homology sequence searching methods such as BLAST  or motif searching methods such as MEME  are not suitable for detecting piRNAs because sequence conservation is very low and no conserved structural motif has been detected in piRNAs.
The first de novo algorithm to identify piRNAs was a position-specific usage method that classifies piRNA sites along the genome using piRNAs starting with a uridine at their 5′ ends. A vector of 21 × 4 components was constructed containing 10 nucleotides upstream and 10 downstream of the starting U (i.e., +10 to −10, where U has the position of 0). The precision of this algorithm was only 61-72%, indicating that this tool is helpful for piRNA classification but still needs improvement . Zhang et al. developed a k-mer based algorithm, named piRNApredictor, to predict piRNAs. piRNA and non-piRNA sequences from five model species were used as the training set. piRNApredictor has a high precision of >90% and a sensitivity of >60% . piRNApredictor was integrated with mirTools 2.0 to predict piRNAs from small RNA-Seq data . Moreover, iMir can be used to find piRNAs , but it mainly focuses on miRNAs. There is another program called "multiclass relevance units machine" that shows an excellent performance on piRNA classification . However, it focuses on algorithm development and its software is not publicly available. proTRAC  and piClust  were developed to display known piRNA clusters, but they cannot be used to find new piRNAs.
Here, we present a new program, piRNA annotation (Piano), to predict piRNAs using piRNA-transposon interaction information. A support vector machine (SVM) was used to classify real piRNAs and pseudo piRNAs. Our analysis of Drosophila melanogaster data shows that Piano performs well in piRNA prediction, with over 90% prediction sensitivity, specificity and accuracy. The SVM classifier trained with Drosophila piRNA data can also accurately identify piRNAs of other species such as Homo sapiens, Mus musculus and Rattus norvegicus. Using small RNA-Seq data, Piano was successfully used to predict piRNAs for an important rice pest, the rice striped stem borer, Chilo suppressalis.
Training and testing sets
Two datasets were built for D. melanogaster: one contained real piRNAs and the other contained pseudo piRNAs. We downloaded 987 piRNAs from the NCBI GenBank database (GI: 157361675–157362817)  and 12,903 piRNAs from the NCBI Gene Expression Omnibus with the accession number GSE9138 . By using short sequence alignment software, SeqMap , highly similar sequences were removed. After removing redundancy, 13,848 non-redundant piRNAs were kept. We downloaded 261,500 Drosophila transposons from the UCSC Genome Browser (Apr. 2006 dm3) . We aligned 13,848 piRNAs to the transposon sequences using SeqMap with a maximum of three mismatches allowed. Among 13,848 non-redundant piRNAs, 9,758 (70.4%) could be aligned successfully, suggesting that they can target transposons.
Since DNA sequences are not random sequences, there are some differences between coding and non-coding RNAs. Because piRNAs are non-coding RNAs, we used non-coding RNAs as a negative control to generate our pseudo piRNA dataset. We downloaded 102,655 Drosophila ncRNA sequences from the NONCODE v3.0 database . First, we removed all piRNAs from this dataset. We then randomly selected one ncRNA sequence and randomly cut out a short sequence of 20–30 nt as one candidate sequence. By this double-randomization process, we were able to obtain about 200,000 candidate pseudo piRNAs. Next, we mapped all these candidate sequences to the transposons with a maximum of three mismatches, and those sequences that did not map to the transposons were removed from the candidate sequence dataset. Accordingly, we produced 38,919 non-redundant candidate pseudo piRNAs. We then randomly selected some candidate pseudo piRNA sequences to simulate the length distribution of real piRNAs. Finally, we obtained 9,240 sequences that formed the pseudo piRNA dataset as the negative dataset for SVM classification.
Cross-species test set
We applied the SVM classifier trained with Drosophila piRNAs to human, mouse and rat data. In total, 32,152 human, 75,814 mouse and 66,758 rat piRNAs were downloaded from the NONCODE v3.0 database . Transposons of the three species were downloaded from the UCSC Genome Browser , including 8,537,572 human, 7,320,714 mouse and 6,380,192 rat transposons.
Structure-sequence triplet elements
Support vector machine
Support vector machines (SVMs) have been widely applied in the classification of biological signals. For a given dataset, x i ∈ R n (i = 1,…N) with corresponding labels y i (y i = +1 or −1, representing real and pseudo piRNAs respectively in this work), SVM gives a decision function , where α i represents the coefficients to be learnt and K is the kernel function. The LibSVM3.12 package (http://www.csie.ntu.edu.tw/~cjlin/libsvm/)  was used to perform the analysis. For optimizing the SVM classifier, the penalty parameter C and the RBF kernel parameter γ were adjusted using the grid search strategy in LibSVM.
Prediction system assessment
Results and discussion
Datasets for SVM classification
For each feature x j , j = 1, …, N, we calculated the mean () and standard deviation () using positive or negative examples, respectively. The results demonstrated that “…G”, “(.(G”, “..(C”, “..(G”, and “(..C” are the top five discriminative elements. Four of them contain continuously unpaired nucleotides, suggesting that binding stability between piRNA-transposon interactions is the key information in classifying real and pseudo piRNAs (Additional file 2: Table S1).
Application of Piano to other species
Cross-species validation results
The high accuracy in predicting mammalian piRNAs achieved by the SVM classifier trained with Drosophila piRNAs suggests that the structure-sequence triplet element represents a conserved feature for piRNAs.
Comparison with other methods
Comparison between results from Piano and piRNApredictor
Prediction of Chilo suppressalispiRNAs
piRNA target sequences
In this study, we developed a novel program for piRNA annotation called Piano. The program uses piRNA-transposon alignment/pairing and piRNA nucleotide content information (i.e., structure-sequence triplet elements) and achieves a high sensitivity, specificity and accuracy of over 90%. To the best of our knowledge, this is the best prediction performance achieved in comparison with other tools, such as piRNApredictor. Piano can be used not only for large-scale piRNA prediction from small RNA sequencing data but also for genome-wide annotation of piRNAs.
The authors thank Dr. Tao He, Junping Zhang and Fei Ma for critical discussions.
- Claverie J-M: Fewer genes, more noncoding RNA. Science. 2005, 309: 1529-1530. 10.1126/science.1116800.View ArticlePubMedGoogle Scholar
- Mattick JS: The functional genomics of noncoding RNA. Science. 2005, 309: 1527-1528. 10.1126/science.1117806.View ArticlePubMedGoogle Scholar
- Xie C, Yuan J, Li H, Li M, Zhao G, Bu D, Zhu W, Wu W, Chen R, Zhao Y: NONCODEv4: exploring the world of long non-coding RNA genes. Nucleic Acids Res. 2014, 42: D98-D103. 10.1093/nar/gkt1222.View ArticlePubMed CentralPubMedGoogle Scholar
- Kutter C, Svoboda P: miRNA, siRNA, piRNA. RNA Biol. 2008, 5: 181-188. 10.4161/rna.7227.View ArticlePubMedGoogle Scholar
- Girard A, Sachidanandam R, Hannon GJ, Carmell MA: A germline-specific class of small RNAs binds mammalian Piwi proteins. Nature. 2006, 442: 199-202.PubMedGoogle Scholar
- Grivna ST, Beyret E, Wang Z, Lin H: A novel class of small RNAs in mouse spermatogenic cells. Genes Dev. 2006, 20: 1709-1714. 10.1101/gad.1434406.View ArticlePubMed CentralPubMedGoogle Scholar
- Houwing S, Kamminga LM, Berezikov E, Cronembold D, Girard A, Van Den Elst H, Filippov DV, Blaser H, Raz E, Moens CB: A role for Piwi and piRNAs in germ cell maintenance and transposon silencing in Zebrafish. Cell. 2007, 129: 69-82. 10.1016/j.cell.2007.03.026.View ArticlePubMedGoogle Scholar
- Kennedy D: Breakthrough of the Year. Science. 2006, 314: 5807-Google Scholar
- Brennecke J, Aravin AA, Stark A, Dus M, Kellis M, Sachidanandam R, Hannon GJ: Discrete small RNA-generating loci as master regulators of transposon activity in Drosophila . Cell. 2007, 128: 1089-1103. 10.1016/j.cell.2007.01.043.View ArticlePubMedGoogle Scholar
- Thomson T, Lin H: The biogenesis and function PIWI proteins and piRNAs: progress and prospect. Annu Rev Cell Dev Biol. 2009, 25: 355-10.1146/annurev.cellbio.24.110707.175327.View ArticlePubMed CentralPubMedGoogle Scholar
- Juliano C, Wang J, Lin H: Uniting germline and stem cells: the function of Piwi proteins and the piRNA pathway in diverse organisms. Annu Rev Genet. 2011, 45: 447-469. 10.1146/annurev-genet-110410-132541.View ArticlePubMedGoogle Scholar
- Lukic S, Chen K: Human piRNAs are under selection in Africans and repress transposable elements. Mol Biol Evol 2011, 28:3061–3067.,
- Lau NC, Seto AG, Kim J, Kuramochi-Miyagawa S, Nakano T, Bartel DP, Kingston RE: Characterization of the piRNA complex from rat testes. Science. 2006, 313: 363-367. 10.1126/science.1130164.View ArticlePubMedGoogle Scholar
- Yin H, Lin H: An epigenetic activation role of Piwi and a Piwi-associated piRNA in Drosophila melanogaster . Nature. 2007, 450: 304-308. 10.1038/nature06263.View ArticlePubMedGoogle Scholar
- Altschul SF, Gish W, Miller W, Myers EW, Lipman DJ: Basic local alignment search tool. J Mol Biol. 1990, 215: 403-410. 10.1016/S0022-2836(05)80360-2.View ArticlePubMedGoogle Scholar
- Bailey TL, Elkan C: Fitting a mixture model by expectation maximization to discover motifs in bipolymers. 1994, University of California, San Diego, Department of Computer Science and EngineeringGoogle Scholar
- Betel D, Sheridan R, Marks DS, Sander C: Computational analysis of mouse piRNA sequence and biogenesis. PLoS Comput Biol. 2007, 3: e222-10.1371/journal.pcbi.0030222.View ArticlePubMed CentralPubMedGoogle Scholar
- Zhang Y, Wang X, Kang L: A k-mer scheme to predict piRNAs and characterize locust piRNAs. Bioinformatics. 2011, 27: 771-776. 10.1093/bioinformatics/btr016.View ArticlePubMed CentralPubMedGoogle Scholar
- Wu J, Liu Q, Wang X, Zheng J, Wang T, You M, Sun ZS, Shi Q: mirTools 2.0 for non-coding RNA discovery, profiling, and functional annotation based on high-throughput sequencing. RNA Biol. 2013, 10: 1087-1092. 10.4161/rna.25193.View ArticlePubMed CentralPubMedGoogle Scholar
- Giurato G, De Filippo MR, Rinaldi A, Hashim A, Nassa G, Ravo M, Rizzo F, Tarallo R, Weisz A: iMir: An integrated pipeline for high-throughput analysis of small non-coding RNA data obtained by smallRNA-Seq. BMC bioinformatics. 2013, 14: 362-10.1186/1471-2105-14-362.View ArticlePubMed CentralPubMedGoogle Scholar
- Menor M, Baek K, Poisson G: Multiclass relevance units machine: benchmark evaluation and application to small ncRNA discovery. BMC Genomics. 2013, 14: S6-10.1186/1471-2164-14-S2-S6.View ArticlePubMed CentralPubMedGoogle Scholar
- Rosenkranz D, Zischler H: proTRAC-a software for probabilistic piRNA cluster detection, visualization and analysis. BMC bioinformatics. 2012, 13: 5-10.1186/1471-2105-13-5.View ArticlePubMed CentralPubMedGoogle Scholar
- Jung I, Park JC, Kim S: piClust: a density based piRNA clustering algorithm. Comput Biol Chem 2014, 50:60–67.,
- Nishida KM, Saito K, Mori T, Kawamura Y, Nagami-Okada T, Inagaki S, Siomi H, Siomi MC: Gene silencing mechanisms mediated by Aubergine–piRNA complexes in Drosophila male gonad. RNA. 2007, 13: 1911-1922. 10.1261/rna.744307.View ArticlePubMed CentralPubMedGoogle Scholar
- Jiang H, Wong WH: SeqMap: mapping massive amount of oligonucleotides to the genome. Bioinformatics. 2008, 24: 2395-2396. 10.1093/bioinformatics/btn429.View ArticlePubMed CentralPubMedGoogle Scholar
- Karolchik D, Barber GP, Casper J, Clawson H, Cline MS, Diekhans M, Dreszer TR, Fujita PA, Guruvadoo L, Haeussler M: The UCSC genome browser database: 2014 update. Nucleic Acids Res. 2014, 42: D764-D770. 10.1093/nar/gkt1168.View ArticlePubMed CentralPubMedGoogle Scholar
- Bu D, Yu K, Sun S, Xie C, Skogerbo G, Miao R, Xiao H, Liao Q, Luo H, Zhao G Zhao H, Liu Z, Liu C, Chen R, Zhao Y: NONCODE v3. 0: integrative annotation of long noncoding RNAs. Nucleic Acids Res 2012, 40:D210–215.,
- Tafer H, Hofacker IL: RNAplex: a fast tool for RNA–RNA interaction search. Bioinformatics. 2008, 24: 2657-2663. 10.1093/bioinformatics/btn193.View ArticlePubMedGoogle Scholar
- Chang C-C, Lin C-J: LIBSVM: a library for support vector machines. ACM Transactions on Intelligent Systems and Technology (TIST). 2011, 2: 27-Google Scholar
- Xue C, Li F, He T, Liu G-P, Li Y, Zhang X: Classification of real and pseudo microRNA precursors using local structure-sequence features and support vector machine. BMC bioinformatics. 2005, 6: 310-10.1186/1471-2105-6-310.View ArticlePubMed CentralPubMedGoogle Scholar
- Dror G, Sorek R, Shamir R: Accurate identification of alternatively spliced exons using support vector machine. Bioinformatics. 2005, 21: 897-901. 10.1093/bioinformatics/bti132.View ArticlePubMedGoogle Scholar
- Golub TR, Slonim DK, Tamayo P, Huard C, Gaasenbeek M, Mesirov JP, Coller H, Loh ML, Downing JR, Caligiuri MA: Molecular classification of cancer: class discovery and class prediction by gene expression monitoring. Science. 1999, 286: 531-537. 10.1126/science.286.5439.531.View ArticlePubMedGoogle Scholar
- Agarwal S, Graepel T, Herbrich R, Har-Peled S, Roth D: Generalization bounds for the area under the ROC curve. Journal of Machine Learning Research. 2005, 393-425.Google Scholar
- Huang CR, Burns KH, Boeke JD: Active transposition in genomes. Annu Rev Genet. 2012, 46: 651-675. 10.1146/annurev-genet-110711-155616.View ArticlePubMed CentralPubMedGoogle Scholar
- Smit AF: Interspersed repeats and other mementos of transposable elements in mammalian genomes. Curr Opin Genet Dev. 1999, 9: 657-663. 10.1016/S0959-437X(99)00031-3.View ArticlePubMedGoogle Scholar
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/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.