A computational method for predicting regulation of human microRNAs on the influenza virus genome
© Zhang et al.; licensee BioMed Central Ltd. 2013
Published: 14 October 2013
While it has been suggested that host microRNAs (miRNAs) may downregulate viral gene expression as an antiviral defense mechanism, such a mechanism has not been explored in the influenza virus for human flu studies. As it is difficult to conduct related experiments on humans, computational studies can provide some insight. Although many computational tools have been designed for miRNA target prediction, there is a need for cross-species prediction, especially for predicting viral targets of human miRNAs. However, finding putative human miRNAs targeting influenza virus genome is still challenging.
We developed machine-learning features and conducted comprehensive data training for predicting interactions between H1N1 genome segments and host miRNA. We defined our seed region as the first ten nucleotides from the 5' end of the miRNA to the 3' end of the miRNA and integrated various features including the number of consecutive matching bases in the seed region of 10 bases, a triplet feature in seed regions, thermodynamic energy, penalty of bulges and wobbles at binding sites, and the secondary structure of viral RNA for the prediction.
Compared to general predictive models, our model fully takes into account the conservation patterns and features of viral RNA secondary structures, and greatly improves the prediction accuracy. Our model identified some key miRNAs including hsa-miR-489, hsa-miR-325, hsa-miR-876-3p and hsa-miR-2117, which target HA, PB2, MP and NS of H1N1, respectively. Our study provided an interesting hypothesis concerning the miRNA-based antiviral defense mechanism against influenza virus in human, i.e., the binding between human miRNA and viral RNAs may not result in gene silencing but rather may block the viral RNA replication.
Influenza is an infectious disease caused by RNA influenza viruses in the family orthomyxoviridae. Influenza viruses are classified into three types: A, B and C. Influenza A infects a wide variety of avian and mammalian species including humans, which can be subdivided into different serotypes based on the antibody response to these viruses . Influenza B virus almost exclusively infects humans, and it has only one known subtype and is less common than influenza A. Influenza C virus can cause a mild upper respiratory disease [2, 3], but it is rare. Influenza A and B genomes each contain eight segments of single-stranded RNA, and C contains seven segments of single-stranded RNA. Each RNA segment encodes one or two proteins . Take influenza A for example; the eight RNA segments are HA (hemagglutinin), NA (neuraminidase), NP (nucleoprotein), M (matrix protein), NS (nonstructural protein), PA (polymerase A), PB1 (polymerase B1), and PB2 (polymerase B2), altogether coding 11 proteins. Influenza virus genome is prone to gene reassortment. The novel influenza, A/H1N1, is a mixed strain  first reported in Mexico and the United States (March 2009), and soon spread over several other nations in 2009.
Influenza viruses only replicate within living cells, and they deliver their genes and accessory proteins into the host cells . Host cells do not passively accept viral infection, but trigger resistance and neutralization actively. Studying the interaction between viruses and host cells is important for understanding the mechanism of pathogenicity so as to search for an appropriate anti-virus method. Recent studies demonstrate miRNAs encoded by viruses or humans may exert an important influence on the interaction between virus and host [7, 8].
Unlike many plant miRNA targets, which are almost completely complementary in open reading frames (ORFs) , the binding between animal miRNAs and their target sites has incomplete complementarity in base-pairing, and binding sites can be found in 3' UTRs, 5' UTRs and coding regions of target genes [19–21]. Experimental miRNA-recognition methods are laborious and time-consuming; hence these methods cannot achieve high throughput currently. Therefore, numerous miRNA target prediction methods have been proposed such as miRanda , TargetScan(S) , RNA22 , Diana-MicroT , PicTar , RNAhybrid , and miTarget , based on seed complementarity, thermodynamics, conservation, Bayesian statistics, SVM, HMM, artificial neural networks, etc. However, available methods suffer from the lack of gold standards of negative examples to build an effective classifier and can hardly make a good balance between high sensitivity and high specificity, which leads to high false positive and false negative rates. Current prediction algorithms lack consistency when compared to each other, and none of the existing prediction tools have been able to incorporate comprehensive features efficiently. While computational analysis of miRNA-mediated antiviral defense has been conducted , no available software tool to predict cross-species miRNA-mediation mechanism has been available until now. It is likely that some special characteristics exist for binding between human miRNA and viral RNA. In particular, influenza viruses mostly have negative-sense single-strand RNAs. In this study, we focused on interaction between human miRNA and viral negative-sense RNAs, which may prevent the viral RNAs from replication and possibly lead to RNA degradation. Such a binding may have different features from interaction between human miRNA and human mRNA, which results in gene silencing through translational repression or target degradation.
In this study, we developed an influenza virus-based multi-level scoring neural network model to predict human miRNAs that may target influenza RNAs. Our model combines viral genome characteristics, RNA secondary structure characteristics, genetic conservative characteristics, and interaction features at seed regions, which work together to greatly improve prediction accuracy and search speed. A hypothesis is proposed for the interaction between the human miRNA and viral negative-sense RNA. Our study may help find a new approach for the prevention and control of the influenza virus.
In this section we conducted seed region feature analyses and compared our method's performance with the other five prevailing prediction algorithms, using a completely independent test data set.
10-nt sequence base-pairing value in seed region of the binding site
N3 statistical information in seed region
The complexity of miRNA-RNA interactions may lead to an inefficient search for miRNA-RNA sequence matching in the miRNA target recognition, as the current search is often based on minimum free energy (MFE) of miRNA:target duplexes. To further reduce the search time, we developed a statistical energy formula for constructing the triplet (N3) feature to represent local MFE concept at the seed region. N3 is produced by consecutive three-base pairings between miRNA and RNA in the seed region. This is a novel feature of miRNA-target base pairing, in contrast to traditional thermodynamic parameters. There are 216 types of triplet base-pairings according to their MFEs. First, based on the statistical energy formula and experimental data, we calculated their MFEs and mapped the results into discrete consecutive integers from 1 to 27 using the formula Mapping Function: , where and if , . The MFE of a triplet is calculated based on dimer thermodynamic parameters. The SCORE(x) discretizes triplet base-pairing scores, as shown in Additional File 1. As an example, with regard to the seed region instance in Figure 2, the triplet feature value is (18, 13, 12, 11, 11, 0, 0, 0) with 8 dimensions.
Gap penalty function calculation of binding sites
Gap penalty of bulge in relationship with gap size and starting position
Starting position of bulge (from 5' end of miRNA)
In Figure 3, the model MSE trend shows that the new model converged slower, but reached a lower MSE value than the control model. This indicates that the introduction of new features affects convergence of the model. More features used in the new model may result in longer convergence iterations, but better results. Figure 4 shows after adding new features, the correlation between the objective values and the model output values improves significantly in both test and validation sets over the control models.
It is worth mentioning that the new target genes feature an extraction method, which can be easily integrated into the traditional target gene model without significant additional computing time.
Performance on completely independent test data
A comparison among different methods in their ACA values
Diana MicroT 3.0
Predicted human miRNAs that regulate the influenza virus genome
Start position of binding and binding energy
Start Position of Binding
In this section we discuss parameter selection rules and seed region features. We also discuss our key finding and its potential biological implication.
H1N1 segments and genome sequences selection
Given the differences in quantities, regions, acquisition methods and data completeness of H1N1 genome sequences in NCBI (http://www.ncbi.nlm.nih.gov/), we could not choose all possible H1N1 genome sequences. It would have led to biases in the result. For H1N1 genome sequences, only a certain number of sequences of each segment per year from the past 13 years (2000-2012) were selected as non-redundant representatives. In order to identify what number of sequences was the most suitable, we chose 10, 20, 30, 40, 50, 70, 90, 100, and 200 sequences per year to perform comparative analyses. The result indicated that 50 sequences were the best [final data sets selected from this process are in Additional Files 2-10.
It has not been proven that human miRNAs could play any direct role in any of the segments of influenza genome. To simplify our research, we chose HA, PB2, MP and NS segments of H1N1 from 2000 to 2012 for our study, since these four influenza virus proteins play essential roles in influenza virus's pathogenicity and infectiousness [30–34]. For example, the strength of the virulence is directly linked to whether HA can be cleaved to HA1 and HA2. NS plays a regulatory role in viral transcription and replication process. PB2 generates the primer required for viral RNA transcription. MP contains matrix proteins and coding proteins (m1, m2, and m3).
H1N1 segments secondary structure and MFE calculation simplification
In our model we used the different scores for the test to determine which step had significant effect on the result. The test confirmed that RNA secondary structure is an important factor for discovering human encoded miRNAs that regulate the influenza virus genome. This is consistent with our previous study , which showed that local RNA structure had a much stronger effect than a global one on the miRNA-RNA binding.
In this study, cross-hybridized binding was considered. We assumed that when an miRNA targets many segments, it loses specificity and its biological effect to inhibit viral RNA will be substantially reduced. Hence, for one miRNA targeting multiple RNAs, we lowered the miRNA's score through dividing by the number of targets. For multiple miRNAs targeting one RNA, we assumed they added more effectiveness for inhibiting the viral RNA. This might be a mechanism of human endogenous miRNAs to improve strengths in targeting influenza virus. So we took multiple miRNAs targeting one RNA into account in our study.
Our result showed that the predicted binding mode between human miRNA and viral negative-sense single-strand RNAs are significantly different from the observed binding mode between human miRNA and human mRNA. In particular, the former has a consecutive 10-nt fully complementary sequence pattern while the latter has 7, 8 or 9-nt complementary sequence pattern. The A:U and G:C ratio in seed, up-stream and down-stream regions, are also different between the two cases. This indicates that the binding between human miRNA and viral RNAs may be much tighter than that between human miRNA and human mRNA. This also suggests that the binding between human miRNA and viral RNAs may not result in gene silencing through translational repression or target degradation, but rather may prevent the viral RNA replication by forming double-strand RNAs between human miRNA and viral negative-sense single-strand RNAs.
In conclusion, we developed a novel model for cross-species miRNA target prediction based on machine learning approach. Compared to general predictive models, our model fully takes into account the conservation patterns and features of viral RNA secondary structures, and greatly improves the prediction accuracy. Using our model, we discovered human encoded miRNAs hsa-miR-489, hsa-miR-325, hsa-miR-876-3p and hsa-miR-2117 targeting HA, PB2, MP and NS of influenza A, respectively. This number of candidates was very small, and thus the results can be used as a basis for biological reverse genetics test experiments for verification. Moreover, next-generation sequencing can also be used to test the effectiveness of our method and our biological hypothesis. In future work, we will extend our study from the four segments to all eight segments. Different score's weight will be considered by adding more H1N1 segment characteristics.
Materials and methods
Three of human encoded miRNA sequences used in this study as examples
We obtained a consensus secondary structure using RNAalifold of the Vienna RNA package (http://rna.tbi.univie.ac.at/cgi-bin/RNAalifold.cgi) with the default parameters (new RNAalifold with RIBOSUM scoring; fold algorithms and basic options: minimum free energy (MFE) and partition function, and avoiding isolated base pairs). We then aligned all the 50 influenza sequences using ClustalW2 with the default parameters (ALINMENT: full; SCORE TYPE: percent; NO END GAPS: yes; ITERATION: none; NUMITER: 1; OUTPUT FORMAT: aln w/numbers; OUTPUT ORDER: aligned; TREE TYPE: none; CORRECT DIST: off; IGNORE GAPS: off; CLUSTERING: NJ).
Artificial neural network design
An artificial neural network was used for the miRNA target prediction, with a suitable selection and representation of the binding site features used as input features. In this study, in addition to the traditional binding site features, we proposed three novel features which have been described.
Sample positive and negative target genes data as a training model of the data set, from positive target genes (788 groups) and negative target genes (4000 groups) randomly, with total of 4788 target genes. The remaining 1200 groups (including 200 positive target genes and 1000 negative target genes) form a model test data set.
Perform feature selection for model training and extract selected features accordingly in the form of vectors, which are applied to the model training.
In order to achieve good convergence of the training, the maximum number of iterations of the neural network training parameters is set to 2000; the MSE difference threshold between two iterations is set at 0.005, and other parameters adopt the default settings.
Test the current training model by the classification accuracy, correlation test, etc.
Repeat steps 1) to 4) 10 times and compare results. Filter out the outlier models that are extraordinarily good (to avoid over-training) and those that are particularly poor. The remaining models should be used to measure the test performance.
N3 statistical information in seed region
Build a target gene prediction model based on multi-feature fusion machine learning
This paper presents an efficient miRNA target prediction approach based on artificial neural networks trained on both positive and negative data as described above. We ranked all the features including 3-mer matching, penalty assessment of binding sites and alignment feature function at binding sites. Through the feature-ranking test, our new features turned out to be non-redundant with higher scores than features used in the traditional methods. Our method can distinguish all six known types of miRNA-target interactions (7mer1A, 7mer-m8, 8mer, 6mer canonical sites, 3'-supplementary sites, and 3'-compensatory sites).
Discover the candidates using our model
The method proposed in this paper was based on scoring, and the secondary structure of RNAs were also considered as an important factor. Viral RNA structure has been demonstrated to be crucial for the adaptability of viruses . For this reason, scoring based on secondary structure was considered in the method. We focused on the bind sites in the stems of RNA secondary structure. Because the stem is stable, we believe that if the bind sites are in the stem region, miRNA will perturb the RNAs more strongly. The secondary structure of any single sequence may not be representative; as a result, the consensus nucleotide sequence was used.
Based on the complementary sites: A sliding window method is applied to search for complete complementary fragments.
Based on the secondary structure of the complementary sites: If the nucleotide was in the stem region, additional reward score is given.
Based on sequence conservation: If nucleotides at the bind site are conserved or nearly conserved across virus strands in different years, additional reward score is given.
The weights of the above three factors were trained empirically. By combining all score components, we ranked the composite scores in ascending order. The miRNA with the highest composite score and the target subsequence were used. We then used RNAfold of Vienna RNA package to get its binding energy. If more than one miRNA had the same score, the one with the lowest binding energy was used for the final result.
"Human encoded miRNAs that regulate the influenza virus genome" A preliminary version of this paper was published in the proceedings of Proceedings of 2012 IEEE 6th International Conference on Systems Biology (ISB), and the current version has extended more than 40% new content compared with our conference paper.
The publication of this work was supported by the National Natural Science Foundation of China (60673099,60873146, and 60971089) and Jilin University Natural Science Foundation (2010).
This article has been published as part of BMC Systems Biology Volume 7 Supplement 2, 2013: Selected articles from The 6th International Conference of Computational Biology. The full contents of the supplement are available online at http://www.biomedcentral.com/bmcsystbiol/supplements/7/S2.
- Hay AJ, Gregory V, Douglas AR, Lin YP: The evolution of human influenza viruses. Philosophical transactions of the Royal Society of London Series B, Biological sciences. 2001, 356 (1416): 1861-1870.PubMed CentralView ArticlePubMedGoogle Scholar
- Katagiri S, Ohizumi A, Homma M: An outbreak of type C influenza in a children's home. J Infect Dis. 1983, 148 (1): 51-56. 10.1093/infdis/148.1.51.View ArticlePubMedGoogle Scholar
- Arias CF, Escalera-Zamudio M, Soto-Del Rio MD, Cobian-Guemes AG, Isa P, Lopez S: Molecular Anatomy of 2009 Influenza Virus A (H1N1). Archives of Medical Research. 2009, 40 (8): 643-654. 10.1016/j.arcmed.2009.10.007.View ArticlePubMedGoogle Scholar
- Nakajima K: [Influenza virus genome structure and encoded proteins]. Nihon rinsho Japanese journal of clinical medicine. 1997, 55 (10): 2542-2546.PubMedGoogle Scholar
- Novel Swine-Origin Influenza AVIT, Dawood FS, Jain S, Finelli L, Shaw MW, Lindstrom S, Garten RJ, Gubareva LV, Xu X, Bridges CB, et al.: Emergence of a novel swine-origin influenza A (H1N1) virus in humans. The New England journal of medicine. 2009, 360 (25): 2605-2615.View ArticleGoogle Scholar
- Smith AE, Helenius A: How viruses enter animal cells. Science. 2004, 304 (5668): 237-242. 10.1126/science.1094823.View ArticlePubMedGoogle Scholar
- David M: Interferons and MicroRNAs. J Interf Cytok Res. 2010, 30 (11): 825-828. 10.1089/jir.2010.0080.View ArticleGoogle Scholar
- Mahajan VS, Drake A, Chen JZ: Virus-specific host miRNAs: antiviral defenses or promoters of persistent infection?. Trends Immunol. 2009, 30 (1): 1-7. 10.1016/j.it.2008.08.009.PubMed CentralView ArticlePubMedGoogle Scholar
- Bartel DP: MicroRNAs: Genomics, biogenesis, mechanism, and function. Cell. 2004, 116 (2): 281-297. 10.1016/S0092-8674(04)00045-5.View ArticlePubMedGoogle Scholar
- Lewis BP, Burge CB, Bartel DP: Conserved seed pairing, often flanked by adenosines, indicates that thousands of human genes are microRNA targets. Cell. 2005, 120 (1): 15-20. 10.1016/j.cell.2004.12.035.View ArticlePubMedGoogle Scholar
- Lecellier CH, Dunoyer P, Arar K, Lehmann-Che J, Eyquem S, Himber C, Saib A, Voinnet O: A cellular MicroRNA mediates antiviral defense in human cells. Science. 2005, 308 (5721): 557-560. 10.1126/science.1108784.View ArticlePubMedGoogle Scholar
- Murakami Y, Aly HH, Tajima A, Inoue I, Shimotohno K: Regulation of the hepatitis C virus genome replication by miR-199a. J Hepatol. 2009, 50 (3): 453-460. 10.1016/j.jhep.2008.06.010.View ArticlePubMedGoogle Scholar
- Hariharan M, Scaria V, Pillai B, Brahmachari SK: Targets for human encoded microRNAs in HIV genes. Biochemical and biophysical research communications. 2005, 337 (4): 1214-1218. 10.1016/j.bbrc.2005.09.183.View ArticlePubMedGoogle Scholar
- Houzet L, Klase Z, Yeung ML, Wu A, Le SY, Quinones M, Jeang KT: The extent of sequence complementarity correlates with the potency of cellular miRNA-mediated restriction of HIV-1. Nucleic Acids Res. 2012Google Scholar
- Heiss BL, Maximova OA, Thach DC, Speicher JM, Pletnev AG: MicroRNA targeting of neurotropic flavivirus: effective control of virus escape and reversion to neurovirulent phenotype. J Virol. 2012, 86 (10): 5647-5659. 10.1128/JVI.07125-11.PubMed CentralView ArticlePubMedGoogle Scholar
- Scaria V, Hariharan M, Maiti S, Pillai B, Brahmachari SK: Host-virus interaction: a new role for microRNAs. Retrovirology. 2006, 3: 68-10.1186/1742-4690-3-68.PubMed CentralView ArticlePubMedGoogle Scholar
- Li Y, Chan EY, Li J, Ni C, Peng X, Rosenzweig E, Tumpey TM, Katze MG: MicroRNA expression and virulence in pandemic influenza virus-infected mice. J Virol. 2010, 84 (6): 3023-3032. 10.1128/JVI.02203-09.PubMed CentralView ArticlePubMedGoogle Scholar
- Voinnet O: Origin, Biogenesis, and Activity of Plant MicroRNAs. Cell. 2009, 136 (4): 669-687. 10.1016/j.cell.2009.01.046.View ArticlePubMedGoogle Scholar
- Tay Y, Zhang JQ, Thomson AM, Lim B, Rigoutsos I: MicroRNAs to Nanog, Oct4 and Sox2 coding regions modulate embryonic stem cell differentiation. Nature. 2008, 455 (7216): 1124-U1112. 10.1038/nature07299.View ArticlePubMedGoogle Scholar
- JJ F, A L-M, HA C: A search for conserved sequences in coding regions reveals that the let-7 microRNA targets Dicer within its coding sequence. Proc Natl Acad Sci USA. 2008, 105 (39): 14879-14884. 10.1073/pnas.0803230105.View ArticleGoogle Scholar
- Lytle JR, Yario TA, Steitz JA: Target mRNAs are repressed as efficiently by microRNA-binding sites in the 5 ' UTR as in the 3 ' UTR. Proceedings of the National Academy of Sciences of the United States of America. 2007, 104 (23): 9667-9672. 10.1073/pnas.0703820104.PubMed CentralView ArticlePubMedGoogle Scholar
- Enright AJ, John B, Gaul U, Tuschl T, Sander C, Marks DS: MicroRNA targets in Drosophila. Genome biology. 2003, 5 (1): R1-10.1186/gb-2003-5-1-r1.PubMed CentralView ArticlePubMedGoogle Scholar
- Lewis BP, Shih IH, Jones-Rhoades MW, Bartel DP, Burge CB: Prediction of mammalian microRNA targets. Cell. 2003, 115 (7): 787-798. 10.1016/S0092-8674(03)01018-3.View ArticlePubMedGoogle Scholar
- Miranda KC, Huynh T, Tay Y, Ang YS, Tam WL, Thomson AM, Lim B, Rigoutsos I: A pattern-based method for the identification of microRNA binding sites and their corresponding heteroduplexes. Cell. 2006, 126 (6): 1203-1217. 10.1016/j.cell.2006.07.031.View ArticlePubMedGoogle Scholar
- Kiriakidou M, Nelson PT, Kouranov A, Fitziev P, Bouyioukos C, Mourelatos Z, Hatzigeorgiou A: A combined computational-experimental approach predicts human microRNA targets. Genes & Development. 2004, 18 (10): 1165-1178. 10.1101/gad.1184704.View ArticleGoogle Scholar
- Krek A, Grun D, Poy MN, Wolf R, Rosenberg L, Epstein EJ, MacMenamin P, da Piedade I, Gunsalus KC, Stoffel M, et al.: Combinatorial microRNA target predictions. Nat Genet. 2005, 37 (5): 495-500. 10.1038/ng1536.View ArticlePubMedGoogle Scholar
- Rehmsmeier M, Steffen P, Hochsmann M, Giegerich R: Fast and effective prediction of microRNA/target duplexes. Rna-a Publication of the Rna Society. 2004, 10 (10): 1507-1517. 10.1261/rna.5248604.View ArticleGoogle Scholar
- Kim SK, Nam JW, Rhee JK, Lee WJ, Zhang BT: miTarget: microRNA target gene prediction using a support vector machine. BMC bioinformatics. 2006, 7:Google Scholar
- Watanabe Y, Kishi A, Yachie N, Kanai A, Tomita M: Computational analysis of microRNA-mediated antiviral defense in humans. Febs Lett. 2007, 581 (24): 4603-4610. 10.1016/j.febslet.2007.08.049.View ArticlePubMedGoogle Scholar
- Yanguez E, Rodriguez P, Goodfellow I, Nieto A: Influenza virus polymerase confers independence of the cellular cap-binding factor eIF4E for viral mRNA translation. Virology. 2012, 422 (2): 297-307. 10.1016/j.virol.2011.10.028.PubMed CentralView ArticlePubMedGoogle Scholar
- Okumura Y, Takahashi E, Yano M, Ohuchi M, Daidoji T, Nakaya T, Bottcher E, Garten W, Klenk HD, Kido H: Novel Type II Transmembrane Serine Proteases, MSPL and TMPRSS13, Proteolytically Activate Membrane Fusion Activity of the Hemagglutinin of Highly Pathogenic Avian Influenza Viruses and Induce Their Multicycle Replication. Journal of Virology. 2010, 84 (10): 5089-5096. 10.1128/JVI.02605-09.PubMed CentralView ArticlePubMedGoogle Scholar
- Ludwig S, Wang XY, Ehrhardt C, Zheng HY, Donelan N, Planz O, Pleschka S, Garcia-Sastre A, Heins G, Wolff T: The influenza A virus NS1 protein inhibits activation of jun N-terminal kinase and AP-1 transcription factors. Journal of Virology. 2002, 76 (21): 11166-11171. 10.1128/JVI.76.21.11166-11171.2002.PubMed CentralView ArticlePubMedGoogle Scholar
- Ginting TE, Shinya K, Kyan Y, Makino A, Matsumoto N, Kaneda S, Kawaoka Y: Amino Acid Changes in Hemagglutinin Contribute to the Replication of Oseltamivir-Resistant H1N1 Influenza Viruses. Journal of Virology. 2012, 86 (1): 121-127. 10.1128/JVI.06085-11.PubMed CentralView ArticlePubMedGoogle Scholar
- Ahn I, Son HS: Evolutionary analysis of human-origin influenza A virus (H3N2) genes associated with the codon usage patterns since 1993. Virus Genes. 2012, 44 (2): 198-206. 10.1007/s11262-011-0687-4.View ArticlePubMedGoogle Scholar
- Zhi Li, Yuanning Liu, Zhang H: RNA Secondary Structure Comparison Based on Dynamic Programming. Journal of Jilin University(Engineering and Technology Edition). 2011, 41 (2): 250-253.Google Scholar
- Griffiths-Jones S, Saini HK, van Dongen S, Enright AJ: miRBase: tools for microRNA genomics. Nucleic Acids Res. 2008, 36 (Database issue): D154-158.PubMed CentralPubMedGoogle Scholar
- Chen T, He HL, Church GM: Modeling gene expression with differential equations. Pacific Symposium on Biocomputing Pacific Symposium on Biocomputing. 1999, 29-40.Google Scholar
- miRbase: the microRNA database. [http://www.mirbase.org/]
- National Center for Biotechnology Information. [http://www.ncbi.nlm.nih.gov/]
- Hao Zhang, Xin Li, Yuaning Liu, Minggang Hu, Zhi Li, Dong Xu: Human encoded miRNAs that regulate the influenza virus genome. Proceedings of 2012 IEEE 6th International Conference on Systems Biology (ISB). 2012, 22-26. IEEE 6th International Conference on Systems Biology (ISB)Google 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/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.