MysiRNA: Improving siRNA efficacy prediction using a machine-learning model combining multi-tools and whole stacking energy (ΔG)

Asymmetric trichotomous partitioning overcomes dataset limitations in building machine learning models for predicting siRNA efficacy

Chemically modified small interfering RNAs (siRNAs) are promising therapeutics guiding sequence-specific silencing of disease genes. Identifying chemically modified siRNA sequences that effectively silence target genes remains challenging. Such determinations necessitate computational algorithms. Machine learning is a powerful predictive approach for tackling biological problems but typically requires datasets significantly larger than most available siRNA datasets. Here, we describe a framework applying machine learning to a small dataset (356 modified sequences) for siRNA efficacy prediction. To overcome noise and biological limitations in siRNA datasets, we apply a trichotomous, two-threshold, partitioning approach, producing several combinations of classification threshold pairs. We then test the effects of different thresholds on random forest machine learning model performance using a novel evaluation metric accounting for class imbalances. We identify thresholds yielding a model with high predictive power, outperforming a linear model generated from the same data, that was predictive upon experimental evaluation. Using a novel model feature extraction method, we observe target site base importances and base preferences consistent with our current understanding of the siRNA-mediated silencing mechanism, with the random forest providing higher resolution than the linear model. This framework applies to any classification challenge involving small biological datasets, providing an opportunity to develop high-performing design algorithms for oligonucleotide therapies.

A Graph Neural Network Approach for the Analysis of siRNA-Target Biological Networks

Many biological systems are characterised by biological entities, as well as their relationships. These interaction networks can be modelled as graphs, with nodes representing bio-entities, such as molecules, and edges representing relations among them, such as interactions. Due to the current availability of a huge amount of biological data, it is very important to consider in silico analysis methods based on, for example, machine learning, that could take advantage of the inner graph structure of the data in order to improve the quality of the results. In this scenario, graph neural networks (GNNs) are recent computational approaches that directly deal with graph-structured data. In this paper, we present a GNN network for the analysis of siRNA-mRNA interaction networks. siRNAs, in fact, are small RNA molecules that are able to bind to target genes and silence them. These events make siRNAs key molecules as RNA interference agents in many biological interaction networks related to severe diseases such as cancer. In particular, our GNN approach allows for the prediction of the siRNA efficacy, which measures the siRNA's ability to bind and silence a gene target. Tested on benchmark datasets, our proposed method overcomes other machine learning algorithms, including the state-of-the-art predictor based on the convolutional neural network, reaching a Pearson correlation coefficient of approximately 73.6%. Finally, we proposed a case study where the efficacy of a set of siRNAs is predicted for a gene of interest. To the best of our knowledge, GNNs were used for the first time in this scenario.

Constructing the boundary between potent and ineffective siRNAs by MG-algorithm with C-features

BACKGROUND

In siRNA based antiviral therapeutics, selection of potent siRNAs is an indispensable step, but these commonly used features are unable to construct the boundary between potent and ineffective siRNAs.

RESULTS

Here, we select potent siRNAs by removing ineffective ones, where these conditions for removals are constructed by C-features of siRNAs, C-features are generated by MG-algorithm, Icc-cluster and the different combinations of some commonly used features, MG-algorithm and Icc-cluster are two different algorithms to search the nearest siRNA neighbors. For the ineffective siRNAs in test data, they are removed from test data by I-iteration, where I-iteration continually updates training data by adding these successively removed siRNAs. Furthermore, the efficacy of siRNAs of test data is predicted by their nearest neighbors of training data.

CONCLUSIONS

By siRNAs of Hencken dataset, results show that our algorithm removes almost ineffective siRNAs from test data, gives the clear boundary between potent and ineffective siRNAs, and accurately predicts the efficacy of siRNAs also. We suggest that our algorithm can provide new insights for selecting the potent siRNAs.

Beyond identity: Understanding the contribution of the 5' nucleotide of the antisense strand to RNAi activity

In both the pharmaceutical and agricultural fields, RNA-based products have capitalized upon the mechanism of RNA interference for targeted reduction of gene expression to improve phenotypes and traits. Reduction in gene expression by RNAi is the result of a small interfering RNA (siRNA) molecule binding to an ARGONAUTE (AGO) protein and directing the effector complex to a homologous region of a target gene's mRNA. siRNAs properties that govern RNA-AGO association have been studied in detail. The siRNA 5' nucleotide (nt) identity has been demonstrated in plants to be an important property responsible for directing association of endogenous small RNAs with different AGO effector proteins. However, it has not been investigated whether the 5' nt identity is an efficacious determinant for topically-applied chemically synthesized siRNAs. In this study, we employed a sandpaper abrasion method to study the silencing efficacies of topically-applied 21 base-pair siRNA duplexes. The MAGNESIUM CHELATASE and GREEN FLUORESCENT PROTEIN genes were selected as endogenous and transgenic gene targets, respectively, to assess the molecular and phenotypic effects of gene silencing. Collections of siRNA variants with different 5' nt identities and different pairing states between the 5' antisense nt and its match in the sense strand of the siRNA duplex were tested for their silencing efficacy. Our results suggest a flexibility in the 5' nt requirement for topically applied siRNA duplexes in planta and highlight the similarity of 5' thermodynamic rules governing topical siRNA efficacy across plants and animals.

Predicting siRNA efficacy based on multiple selective siRNA representations and their combination at score level

Small interfering RNAs (siRNAs) may induce to targeted gene knockdown, and the gene silencing effectiveness relies on the efficacy of the siRNA. Therefore, the task of this paper is to construct an effective siRNA prediction method. In our work, we try to describe siRNA from both quantitative and qualitative aspects. For quantitative analyses, we form four groups of effective features, including nucleotide frequencies, thermodynamic stability profile, thermodynamic of siRNA-mRNA interaction, and mRNA related features, as a new mixed representation, in which thermodynamic of siRNA-mRNA interaction is introduced to siRNA efficacy prediction for the first time to our best knowledge. And then an F-score based feature selection is employed to investigate the contribution of each feature and remove the weak relevant features. Meanwhile, we encode the siRNA sequence and existed empirical design rules as a qualitative siRNA representation. These two kinds of siRNA representations are combined to predict siRNA efficacy by supported Vector Regression (SVR) at score level. The experimental results indicate that our method may select the features with powerful discriminative ability and make the two kinds of siRNA representations work at full capacity. The prediction results also demonstrate that our method can outperform other popular siRNA efficacy prediction algorithms.

SMEpred workbench: A web server for predicting efficacy of chemicallymodified siRNAs

Chemical modifications have been extensively exploited to circumvent shortcomings in therapeutic applications of small interfering RNAs (siRNAs). However, experimental designing and testing of these siRNAs or chemically modified siRNAs (cm-siRNAs) involves enormous resources. Therefore, in-silico intervention in designing cm-siRNAs would be of utmost importance. We developed SMEpred workbench to predict the efficacy of normal siRNAs as well as cm-siRNAs using 3031 heterogeneous cm-siRNA sequences from siRNAmod database. These include 30 frequently used chemical modifications on different positions of either siRNA strand. Support Vector Machine (SVM) was employed to develop predictive models utilizing various sequence features namely mono-, di-nucleotide composition, binary pattern and their hybrids. We achieved highest Pearson Correlation Coefficient (PCC) of 0.80 during 10-fold cross validation and similar PCC value in independent validation. We have provided the algorithm in the 'SMEpred' pipeline to predict the normal siRNAs from the gene or mRNA sequence. For multiple modifications, we have assembled 'MultiModGen' module to design multiple modifications and further process them to evaluate their predicted efficacies. SMEpred webserver will be useful to scientific community engaged in use of RNAi-based technology as well as for therapeutic development. Web server is available for public use at following URL address: http://bioinfo.imtech.res.in/manojk/smepred .

In Silico Design and Experimental Validation of siRNAs Targeting Conserved Regions of Multiple Hepatitis C Virus Genotypes

RNA interference (RNAi) is a post-transcriptional gene silencing mechanism that mediates the sequence-specific degradation of targeted RNA and thus provides a tremendous opportunity for development of oligonucleotide-based drugs. Here, we report on the design and validation of small interfering RNAs (siRNAs) targeting highly conserved regions of the hepatitis C virus (HCV) genome. To aim for therapeutic applications by optimizing the RNAi efficacy and reducing potential side effects, we considered different factors such as target RNA variations, thermodynamics and accessibility of the siRNA and target RNA, and off-target effects. This aim was achieved using an in silico design and selection protocol complemented by an automated MysiRNA-Designer pipeline. The protocol included the design and filtration of siRNAs targeting highly conserved and accessible regions within the HCV internal ribosome entry site, and adjacent core sequences of the viral genome with high-ranking efficacy scores. Off-target analysis excluded siRNAs with potential binding to human mRNAs. Under this strict selection process, two siRNAs (HCV353 and HCV258) were selected based on their predicted high specificity and potency. These siRNAs were tested for antiviral efficacy in HCV genotype 1 and 2 replicon cell lines. Both in silico-designed siRNAs efficiently inhibited HCV RNA replication, even at low concentrations and for short exposure times (24h); they also exceeded the antiviral potencies of reference siRNAs targeting HCV. Furthermore, HCV353 and HCV258 siRNAs also inhibited replication of patient-derived HCV genotype 4 isolates in infected Huh-7 cells. Prolonged treatment of HCV replicon cells with HCV353 did not result in the appearance of escape mutant viruses. Taken together, these results reveal the accuracy and strength of our integrated siRNA design and selection protocols. These protocols could be used to design highly potent and specific RNAi-based therapeutic oligonucleotide interventions.

Modified gold nanoparticles for intracellular delivery of anti-liver cancer siRNA

To overcome the rapid enzymatic degradation and low transfection efficiency of siRNA, the delivery carriers for siRNA is a therapeutic demand to increase its stability. Gold nanoparticles (AuNPs) modified by branched polyethyleneimine (bPEI) were developed as an efficient and safe intracellular delivery carriers for siRNA. The current study implied that siRNA designed against an oncogene c-Myc could be delivered by a modified AuNPs complex without significant cytotoxicity. The comparative semi-quantitative and quantitative real time PCR were used to measure the c-Myc gene expression after transfection with naked siRNA and siRNA/bPEI/AuNPs, but AuNPs interfered with PCR. However, the c-Myc protein translation was successfully detected in the transfected HuH7 cells with naked siRNA and siRNA/bPEI/AuNPs and it was found to be inhibited by siRNA/bPEI/AuNPs more than naked siRNA. The results validate the successful silencing of c-Myc gene. Accordingly, it may confirm the promising and effective delivery of siRNA by bPEI/AuNPs. The complex enhances the cellular uptake of siRNA without significant cytotoxicity and confirms that bPEI modified AuNPs could be used as a good candidate for safe cellular delivery of siRNA.

Soft computing model for optimized siRNA design by identifying off target possibilities using artificial neural network model

The ability of small interfering RNA (siRNA) to do posttranscriptional gene regulation by knocking down targeted genes is an important research topic in functional genomics, biomedical research and in cancer therapeutics. Many tools had been developed to design exogenous siRNA with high experimental inhibition. Even though considerable amount of work has been done in designing exogenous siRNA, design of effective siRNA sequences is still a challenging work because the target mRNAs must be selected such that their corresponding siRNAs are likely to be efficient against that target and unlikely to accidentally silence other transcripts due to sequence similarity. In some cases, siRNAs may tolerate mismatches with the target mRNA, but knockdown of genes other than the intended target could make serious consequences. Hence to design siRNAs, two important concepts must be considered: the ability in knocking down target genes and the off target possibility on any nontarget genes. So before doing gene silencing by siRNAs, it is essential to analyze their off target effects in addition to their inhibition efficacy against a particular target. Only a few methods have been developed by considering both efficacy and off target possibility of siRNA against a gene. In this paper we present a new design of neural network model with whole stacking energy (ΔG) that enables to identify the efficacy and off target effect of siRNAs against target genes. The tool lists all siRNAs against a particular target with their inhibition efficacy and number of matches or sequence similarity with other genes in the database. We could achieve an excellent performance of Pearson Correlation Coefficient (R=0. 74) and Area Under Curve (AUC=0.906) when the threshold of whole stacking energy is ≥-34.6 kcal/mol. To the best of the author's knowledge, this is one of the best score while considering the "combined efficacy and off target possibility" of siRNA for silencing a gene. The proposed model shall be useful for designing exogenous siRNA for therapeutic applications and gene silencing techniques in the area of bioinformatics. The software is developed as a desktop application and available at http://opsid.in/opsid/.

A semi-supervised tensor regression model for siRNA efficacy prediction

BACKGROUND

Short interfering RNAs (siRNAs) can knockdown target genes and thus have an immense impact on biology and pharmacy research. The key question of which siRNAs have high knockdown ability in siRNA research remains challenging as current known results are still far from expectation.

RESULTS

This work aims to develop a generic framework to enhance siRNA knockdown efficacy prediction. The key idea is first to enrich siRNA sequences by incorporating them with rules found for designing effective siRNAs and representing them as enriched matrices, then to employ the bilinear tensor regression to predict knockdown efficacy of those matrices. Experiments show that the proposed method achieves better results than existing models in most cases.

CONCLUSIONS

Our model not only provides a suitable siRNA representation but also can predict siRNA efficacy more accurate and stable than most of state-of-the-art models. Source codes are freely available on the web at: http://www.jaist.ac.jp/\~bao/BiLTR/ .

Influence of mRNA features on siRNA interference efficacy

Design of small interference RNA (siRNA) is one of the most important steps in effectively applying the RNA interference (RNAi) technology. The current siRNA design often produces inconsistent design results, which often fail to reliably select siRNA with clear silencing effects. We propose that when designing siRNA, one should consider mRNA global features and near siRNA-binding site local features. By a linear regression study, we discovered strong correlations between inhibitory efficacy and both mRNA global features and neighboring local features. This paper shows that, on average, less GC content, fewer stem secondary structures, and more loop secondary structures of mRNA at both global and local flanking regions of the siRNA binding sites lead to stronger inhibitory efficacy. Thus, the use of mRNA global features and near siRNA-binding site local features are essential to successful gene silencing and hence, a better siRNA design. We use a random forest model to predict siRNA efficacy using siRNA features, mRNA features, and near siRNA binding site features. Our prediction method achieved a correlation coefficient of 0.7 in 10-fold cross validation in contrast to 0.63 when using siRNA features only. Our study demonstrates that considering mRNA and near siRNA binding site features helps improve siRNA design accuracy. The findings may also be helpful in understanding binding efficacy between microRNA and mRNA.

DSIR: assessing the design of highly potent siRNA by testing a set of cancer-relevant target genes

Chemically synthesized small interfering RNA (siRNA) is a widespread molecular tool used to knock down genes in mammalian cells. However, designing potent siRNA remains challenging. Among tools predicting siRNA efficacy, very few have been validated on endogenous targets in realistic experimental conditions. We previously described a tool to assist efficient siRNA design (DSIR, Designer of siRNA), which focuses on intrinsic features of the siRNA sequence. Here, we evaluated DSIR’s performance by systematically investigating the potency of the siRNA it designs to target ten cancer-related genes. mRNA knockdown was measured by quantitative RT-PCR in cell-based assays, revealing that over 60% of siRNA sequences designed by DSIR silenced their target genes by at least 70%. Silencing efficacy was sustained even when low siRNA concentrations were used. This systematic analysis revealed in particular that, for a subset of genes, the efficiency of siRNA constructs significantly increases when the sequence is located closer to the 5′-end of the target gene coding sequence, suggesting the distance to the 5′-end as a new feature for siRNA potency prediction. A new version of DSIR incorporating these new findings, as well as the list of validated siRNA against the tested cancer genes, has been made available on the web (http://biodev.extra.cea.fr/DSIR).

MysiRNA-designer: a workflow for efficient siRNA design

The design of small interfering RNA (siRNA) is a multi factorial problem that has gained the attention of many researchers in the area of therapeutic and functional genomics. MysiRNA score was previously introduced that improves the correlation of siRNA activity prediction considering state of the art algorithms. In this paper, a new program, MysiRNA-Designer, is described which integrates several factors in an automated work-flow considering mRNA transcripts variations, siRNA and mRNA target accessibility, and both near-perfect and partial off-target matches. It also features the MysiRNA score, a highly ranked correlated siRNA efficacy prediction score for ranking the designed siRNAs, in addition to top scoring models Biopredsi, DISR, Thermocomposition21 and i-Score, and integrates them in a unique siRNA score-filtration technique. This multi-score filtration layer filters siRNA that passes the 90% thresholds calculated from experimental dataset features. MysiRNA-Designer takes an accession, finds conserved regions among its transcript space, finds accessible regions within the mRNA, designs all possible siRNAs for these regions, filters them based on multi-scores thresholds, and then performs SNP and off-target filtration. These strict selection criteria were tested against human genes in which at least one active siRNA was designed from 95.7% of total genes. In addition, when tested against an experimental dataset, MysiRNA-Designer was found capable of rejecting 98% of the false positive siRNAs, showing superiority over three state of the art siRNA design programs. MysiRNA is a freely accessible (Microsoft Windows based) desktop application that can be used to design siRNA with a high accuracy and specificity. We believe that MysiRNA-Designer has the potential to play an important role in this area.