Predicting siRNA activity based on back-propagation neural network

RNA interference (RNAi) is a phenomenon of gene silence induced by a double-stranded RNA (dsRNA) homologous to a target gene.RNAi can be used to identify the function of genes or to knock down the targeted genes.In RNAi technology,19 bp double-stranded short interfering RNAs (siRNA) with characteristic 3' overhangs are usually used.The effects of siRNAs are quite varied due to the different choices in the sites of target mRNA.Moreover,there are many factors influencing siRNA activity and these factors are usually nonlinear.To find the motif features and the effect on siRNA activity,we carried out a feature extraction on some published experimental data and used these features to train a backpropagation neural network (BP NN).Then,we used the trained BP NN to predict siRNA activity.     

Improved and automated prediction of effective siRNA

Short interfering RNAs are used in functional genomics studies to knockdown a single gene in a reversible manner. The results of siRNA experiments are highly dependent on the choice of siRNA sequence. In order to evaluate siRNA design rules, we collected a database of 398 siRNAs of known efficacy from 92 genes. We used this database to evaluate previously proposed rules from smaller datasets, and to find a new set of rules that are optimal for the entire database. We also trained a regression tree with full cross-validation. It was however difficult to obtain the same precision as methods previously tested on small datasets from one or two genes. We show that those methods are overfitting as they work poorly on independent validation datasets from multiple genes. Our new design rules can predict siRNAs with efficacy >/= 50% in 91% of cases, and with efficacy >/=90% in 52% of cases, which is more than a twofold improvement over random selection. Software for designing siRNAs is available online via a web server at or as a standalone version for high-throughput applications.     

siSPOTR: a tool for designing highly specific and potent siRNAs for human and mouse

RNA interference (RNAi) serves as a powerful and widely used gene silencing tool for basic biological research and is being developed as a therapeutic avenue to suppress disease-causing genes. However, the specificity and safety of RNAi strategies remains under scrutiny because small inhibitory RNAs (siRNAs) induce off-target silencing. Currently, the tools available for designing siRNAs are biased toward efficacy as opposed to specificity. Prior work from our laboratory and others’ supports the potential to design highly specific siRNAs by limiting the promiscuity of their seed sequences (positions 2–8 of the small RNA), the primary determinant of off-targeting. Here, a bioinformatic approach to predict off-targeting potentials was established using publically available siRNA data from more than 50 microarray experiments. With this, we developed a specificity-focused siRNA design algorithm and accompanying online tool which, upon validation, identifies candidate sequences with minimal off-targeting potentials and potent silencing capacities. This tool offers researchers unique functionality and output compared with currently available siRNA design programs. Furthermore, this approach can greatly improve genome-wide RNAi libraries and, most notably, provides the only broadly applicable means to limit off-targeting from RNAi expression vectors.     

Genomic analysis of anti-hepatitis B virus (HBV) activity by small interfering RNA and lamivudine in stable HBV-producing cells

Hepatitis B virus (HBV) causes acute and chronic hepatitis and hepatocellular carcinoma. Small interfering RNA (siRNA) and lamivudine have been shown to have anti-HBV effects through different mechanisms. However, assessment of the genome-wide effects of siRNA and lamivudine on HBV-producing cell lines has not been reported, which may provide a clue to interrogate the HBV-cell interaction and to evaluate the siRNA's side effect as a potential drug. In the present study, we designed seven siRNAs based on the conserved HBV sequences and tested their effects on the expression of HBV genes following sorting of siRNA-positive cells. Among these seven siRNAs, siRNA-1 and siRNA-7 were found to effectively suppress HBV gene expression. We further addressed the global gene expression changes in stable HBV-producing cells induced by siRNA-1 and siRNA-7 by use of human genome-wide oligonucleotide microarrays. Data from the gene expression profiling indicated that siRNA-1 and siRNA-7 altered the expression of 54 and 499 genes, respectively, in HepG2.2.15 cells, which revealed that different siRNAs had various patterns of gene expression profiles and suggested a complicated influence of siRNAs on host cells. We further observed that 18 of these genes were suppressed by both siRNA-1 and siRNA-7. Interestingly, seven of these genes were originally activated by HBV, which suggested that these seven genes might be involved in the HBV-host cell interaction. Finally, we have compared the effects of siRNA and lamivudine on HBV and host cells, which revealed that siRNA is more effective at inhibiting HBV expression at the mRNA and protein level in vitro, and the gene expression profile of HepG2.2.15 cells treated by lamivudine is totally different from that seen with siRNA.     

Accurate transcriptome-wide prediction of microRNA targets and small interfering RNA off-targets with MIRZA-G

Small interfering RNA (siRNA)-mediated knock-down is a widely used experimental approach to characterizing gene function. Although siRNAs are designed to guide the cleavage of perfectly complementary mRNA targets, acting similarly to microRNAs (miRNAs), siRNAs down-regulate the expression of hundreds of genes to which they have only partial complementarity. Prediction of these siRNA ‘off-targets’ remains difficult, due to the incomplete understanding of siRNA/miRNA–target interactions. Combining a biophysical model of miRNA–target interaction with structure and sequence features of putative target sites we developed a suite of algorithms, MIRZA-G, for the prediction of miRNA targets and siRNA off-targets on a genome-wide scale. The MIRZA-G variant that uses evolutionary conservation performs better than currently available methods in predicting canonical miRNA target sites and in addition, it predicts non-canonical miRNA target sites with similarly high accuracy. Furthermore, MIRZA-G variants predict siRNA off-target sites with an accuracy unmatched by currently available programs. Thus, MIRZA-G may prove instrumental in the analysis of data resulting from large-scale siRNA screens.     

Competition for RISC binding predicts in vitro potency of siRNA

Short interfering RNAs (siRNA) guide degradation of target RNA by the RNA-induced silencing complex (RISC). The use of siRNA in animals is limited partially due to the short half-life of siRNAs in tissues. Chemically modified siRNAs are necessary that maintain mRNA degradation activity, but are more stable to nucleases. In this study, we utilized alternating 2′-O-methyl and 2′-deoxy-2′-fluoro (OMe/F) chemically modified siRNA targeting PTEN and Eg5. OMe/F-modified siRNA consistently reduced mRNA and protein levels with equal or greater potency and efficacy than unmodified siRNA. We showed that modified siRNAs use the RISC mechanism and lead to cleavage of target mRNA at the same position as unmodified siRNA. We further demonstrated that siRNAs can compete with each other, where highly potent siRNAs can compete with less potent siRNAs, thus limiting the ability of siRNAs with lower potency to mediate mRNA degradation. In contrast, a siRNA with low potency cannot compete with a highly efficient siRNA. We established a correlation between siRNA potency and ability to compete with other siRNAs. Thus, siRNAs that are more potent inhibitors for mRNA destruction have the potential to out-compete less potent siRNAs indicating that the amount of a cellular component, perhaps RISC, limits siRNA activity.     

Use of bicistronic vectors in combination with flow cytometry to screen for effective small interfering RNA target sequences

The efficacy and specificity of small interfering RNAs (siRNAs) are largely dependent on the siRNA sequence. Since only empirical strategies are currently available for predicting these parameters, simple and accurate methods for evaluating siRNAs are needed. To simplify such experiments, target genes are often tagged with reporters for easier readout. Here, we used a bicistronic vector expressing a target gene and green fluorescent protein (GFP) to create a system in which the effect of an siRNA sequence was reflected in the GFP expression level. Cells were transduced with the bicistronic vector, expression vectors for siRNA and red fluorescent protein (RFP). Flow cytometric analysis of the transduced cells revealed that siRNAs for the target gene silenced GFP from the bicistronic vector, but did not silence GFP transcribed without the target gene sequence. In addition, the mean fluorescence intensities of GFP on RFP-expressing cells correlated well with the target gene mRNA and protein levels. These results suggest that this flow cytometry-based method enables us to quantitatively evaluate the efficacy and specificity of siRNAs. Because of its simplicity and effectiveness, this method will facilitate the screening of effective siRNA target sequences, even in high-throughput applications.     

Predicting siRNA activity based on back-propagation neural network

RNA interference (RNAi) is a phenomenon of gene silence induced by a double-stranded RNA (dsRNA) homologous to a target gene. RNAi can be used to identify the function of genes or to knock down the targeted genes. In RNAi technology, 19 bp double-stranded short interfering RNAs (siRNA) with characteristic 39 overhangs are usually used. The effects of siRNAs are quite varied due to the different choices in the sites of target mRNA. Moreover, there are many factors influencing siRNA activity and these factors are usually nonlinear. To find the motif features and the effect on siRNA activity, we carried out a feature extraction on some published experimental data and used these features to train a back-propagation neural network (BP NN). Then, we used the trained BP NN to predict siRNA activity. __________ Translated from Acta Biophysica Sinica, 2006, 22(6): 429–434 [译自: 生物物理学报]     

Prediction of siRNA knockdown efficiency using artificial neural network models

Selective knockdown of gene expression by short interference RNAs (siRNAs) has allowed rapid validation of gene functions and made possible a high throughput, genome scale approach to interrogate gene function. However, randomly designed siRNAs display different knockdown efficiencies of target genes. Hence, various prediction algorithms based on siRNA functionality have recently been constructed to increase the likelihood of selecting effective siRNAs, thereby reducing the experimental cost. Toward this end, we have trained three Back-propagation and Bayesian neural network models, previously not used in this context, to predict the knockdown efficiencies of 180 experimentally verified siRNAs on their corresponding target genes. Using our input coding based primarily on RNA structure thermodynamic parameters and cross-validation method, we showed that our neural network models outperformed most other methods and are comparable to the best predicting algorithm thus far published. Furthermore, our neural network models correctly classified 74% of all siRNAs into different efficiency categories; with a correlation coefficient of 0.43 and receiver operating characteristic curve score of 0.78, thus highlighting the potential utility of this method to complement other existing siRNA classification and prediction schemes.     

Designing of highly effective complementary and mismatch siRNAs for silencing a gene

In past, numerous methods have been developed for predicting efficacy of short interfering RNA (siRNA). However these methods have been developed for predicting efficacy of fully complementary siRNA against a gene. Best of author's knowledge no method has been developed for predicting efficacy of mismatch siRNA against a gene. In this study, a systematic attempt has been made to identify highly effective complementary as well as mismatch siRNAs for silencing a gene.Support vector machine (SVM) based models have been developed for predicting efficacy of siRNAs using composition, binary and hybrid pattern siRNAs. We achieved maximum correlation 0.67 between predicted and actual efficacy of siRNAs using hybrid model. All models were trained and tested on a dataset of 2182 siRNAs and performance was evaluated using five-fold cross validation techniques. The performance of our method desiRm is comparable to other well-known methods. In this study, first time attempt has been made to design mutant siRNAs (mismatch siRNAs). In this approach we mutated a given siRNA on all possible sites/positions with all possible nucleotides. Efficacy of each mutated siRNA is predicted using our method desiRm. It is well known from literature that mismatches between siRNA and target affects the silencing efficacy. Thus we have incorporated the rules derived from base mismatches experimental data to find out over all efficacy of mutated or mismatch siRNAs. Finally we developed a webserver, desiRm (http://www.imtech.res.in/raghava/desirm/) for designing highly effective siRNA for silencing a gene. This tool will be helpful to design siRNA to degrade disease isoform of heterozygous single nucleotide polymorphism gene without depleting the wild type protein.     

Forward and robust selection of the most potent and noncellular toxic siRNAs from RNAi libraries

Use of highly potent small interfering RNAs (siRNAs) can substantially reduce dose-dependent cytotoxic and off-target effects. We developed a genetic forward approach by fusing the cytosine deaminase gene with targets for the robust identification of highly potent siRNAs from RNA interference (RNAi) libraries that were directly delivered into cells via bacterial invasion. We demonstrated that two simple drug selection cycles performed conveniently in a single container predominately enriched two siRNAs targeting the MVP gene (siMVP) and one siRNA targeting the egfp gene (siEGFP) in surviving cells and these proved to be the most effective siRNAs reported. Furthermore, the potent siRNAs isolated from the surviving cells possessed noncellular toxic characteristics. Interestingly, the length of highly potent siMVPs identified could be as short as 16-mer, and increasing the length of their native sequences dramatically reduced RNAi potency. These results suggest that the current approach can robustly discover the most potent and nontoxic siRNAs in the surviving cells, and thus has great potential in facilitating RNAi applications by minimizing the dose-dependent and sequence nonspecific side effects of siRNAs.     

Rational siRNA design for RNA interference

Short-interfering RNAs suppress gene expression through a highly regulated enzyme-mediated process called RNA interference (RNAi). RNAi involves multiple RNA-protein interactions characterized by four major steps: assembly of siRNA with the RNA-induced silencing complex (RISC), activation of the RISC, target recognition and target cleavage. These interactions may bias strand selection during siRNA-RISC assembly and activation, and contribute to the overall efficiency of RNAi. To identify siRNA-specific features likely to contribute to efficient processing at each step, we performed a systematic analysis of 180 siRNAs targeting the mRNA of two genes. Eight characteristics associated with siRNA functionality were identified: low G/C content, a bias towards low internal stability at the sense strand 3'-terminus, lack of inverted repeats, and sense strand base preferences (positions 3, 10, 13 and 19). Further analyses revealed that application of an algorithm incorporating all eight criteria significantly improves potent siRNA selection. This highlights the utility of rational design for selecting potent siRNAs and facilitating functional gene knockdown studies.     

Designing highly active siRNAs for therapeutic applications

The discovery of RNA interference (RNAi) generated considerable interest in developing short interfering RNAs (siRNAs) for understanding basic biology and as the active agents in a new variety of therapeutics. Early studies showed that selecting an active siRNA was not as straightforward as simply picking a sequence on the target mRNA and synthesizing the siRNA complementary to that sequence. As interest in applying RNAi has increased, the methods for identifying active siRNA sequences have evolved from focusing on the simplicity of synthesis and purification, to identifying preferred target sequences and secondary structures, to predicting the thermodynamic stability of the siRNA. As more specific details of the RNAi mechanism have been defined, these have been incorporated into more complex siRNA selection algorithms, increasing the reliability of selecting active siRNAs against a single target. Ultimately, design of the best siRNA therapeutics will require design of the siRNA itself, in addition to design of the vehicle and other components necessary for it to function in vivo. In this minireview, we summarize the evolution of siRNA selection techniques with a particular focus on one issue of current importance to the field, how best to identify those siRNA sequences likely to have high activity. Approaches to designing active siRNAs through chemical and structural modifications will also be highlighted. As the understanding of how to control the activity and specificity of siRNAs improves, the potential utility of siRNAs as human therapeutics will concomitantly grow.