Computational prediction of antisense oligonucleotides and siRNAs
Author: Chalk, Alistair
Date: 2005-09-23
Location: CMB auditorium, Berzius väg 21, Karolinska Institutet
Time: 15.00
Department: Centrum för Genomik och Bioinformatik (CGB) / Center for Genomics Research
Abstract
Two popular gene knockdown methods, antisense oligonucleotides (AOs) and short interfering RNAs (siRNAs) are commonly used to selectively inhibit gene expression (gene knockdown). This is an extremely valuable tool for functional genomics, however the selection of effective molecules using either method is non trivial. Here we present a number of computational solutions including predictive methods for AO and siRNA molecules, a database of siRNAs of known efficacy and a visualization tool for viewing complex sequence analysis scenarios.
Prediction models for AOs based on the machine learning methods Artificial Neural Networks (AOpredict, paper I) and Support Vector Machines (SVMPredict, paper II) address the problem of AO design by applying computational methods to a database of AOs mined from the literature. These models predict AO efficacy at high accuracy according to cross-validation results. A database of siRNAs was created from the literature and contains 1276 siRNAs targeting 116 genes; this database is available through a web interface (siRNAdb, paper III).
The properties of functional and non-functional siRNAs were examined, current siRNA design rules were evaluated and a new prediction method was developed (siSearch, paper IV). The issue of off-target effects was examined in detail and a scoring scheme was developed based on the available experimental data (SpecificityServer, paper V). We identified that 14-23% of siRNAs in the database are likely to elicit off-target effects. A tool for viewing multiple series of biological information and computational predictions was developed to view data from SFS and GFF formats, simplifying the analysis process (Sfixem, paper VI).
In an independent validation of 63 siRNAs designed by siSearch to target 31 Rat genes, 19 (30%) had a silencing activity of 90% or higher and 3 3 (52%) silenced to a level of 70% or more. For 25 of the 31 genes siSearch was successful in predicting siRNAs that silence the gene by more than 80%. The prediction methods and specificity server, all of which are available online are highly applicable tools for die practical design of high efficacy specific siRNAs and AOs.
Prediction models for AOs based on the machine learning methods Artificial Neural Networks (AOpredict, paper I) and Support Vector Machines (SVMPredict, paper II) address the problem of AO design by applying computational methods to a database of AOs mined from the literature. These models predict AO efficacy at high accuracy according to cross-validation results. A database of siRNAs was created from the literature and contains 1276 siRNAs targeting 116 genes; this database is available through a web interface (siRNAdb, paper III).
The properties of functional and non-functional siRNAs were examined, current siRNA design rules were evaluated and a new prediction method was developed (siSearch, paper IV). The issue of off-target effects was examined in detail and a scoring scheme was developed based on the available experimental data (SpecificityServer, paper V). We identified that 14-23% of siRNAs in the database are likely to elicit off-target effects. A tool for viewing multiple series of biological information and computational predictions was developed to view data from SFS and GFF formats, simplifying the analysis process (Sfixem, paper VI).
In an independent validation of 63 siRNAs designed by siSearch to target 31 Rat genes, 19 (30%) had a silencing activity of 90% or higher and 3 3 (52%) silenced to a level of 70% or more. For 25 of the 31 genes siSearch was successful in predicting siRNAs that silence the gene by more than 80%. The prediction methods and specificity server, all of which are available online are highly applicable tools for die practical design of high efficacy specific siRNAs and AOs.
List of papers:
I. Chalk AM, Sonnhammer EL (2002). Computational antisense oligo prediction with a neural network model. Bioinformatics. 18(12): 1567-75.
Pubmed
II. Camps-Valls G, Chalk AM, Serrano-Lopez AJ, Martin-Guerrero JD, Sonnhammer EL (2004). Profiled support vector machines for antisense oligonucleotide efficacy prediction. BMC Bioinformatics. 5(1): 135.
Pubmed
III. Chalk AM, Warfinge RE, Georgii-Hemming P, Sonnhammer EL (2005). siRNAdb: a database of siRNA sequences. Nucleic Acids Res. 33 (Database issue): D131-4.
Pubmed
IV. Chalk AM, Wahlestedt C, Sonnhammer EL (2004). Improved and automated prediction of effective siRNA. Biochem Biophys Res Commun. 319(1): 264-74.
Pubmed
V. Chalk AM, Du Q, Liang Z, Sonnhammer EL (2005). siRNA specificity searching incorporating mismatch tolerance data. [Manuscript]
VI. Chalk AM, Wennerberg M, Sonnhammer EL (2004). Sfixem - graphical sequence feature display in Java. Bioinformatics. 20(15): 2488-90.
Pubmed
I. Chalk AM, Sonnhammer EL (2002). Computational antisense oligo prediction with a neural network model. Bioinformatics. 18(12): 1567-75.
Pubmed
II. Camps-Valls G, Chalk AM, Serrano-Lopez AJ, Martin-Guerrero JD, Sonnhammer EL (2004). Profiled support vector machines for antisense oligonucleotide efficacy prediction. BMC Bioinformatics. 5(1): 135.
Pubmed
III. Chalk AM, Warfinge RE, Georgii-Hemming P, Sonnhammer EL (2005). siRNAdb: a database of siRNA sequences. Nucleic Acids Res. 33 (Database issue): D131-4.
Pubmed
IV. Chalk AM, Wahlestedt C, Sonnhammer EL (2004). Improved and automated prediction of effective siRNA. Biochem Biophys Res Commun. 319(1): 264-74.
Pubmed
V. Chalk AM, Du Q, Liang Z, Sonnhammer EL (2005). siRNA specificity searching incorporating mismatch tolerance data. [Manuscript]
VI. Chalk AM, Wennerberg M, Sonnhammer EL (2004). Sfixem - graphical sequence feature display in Java. Bioinformatics. 20(15): 2488-90.
Pubmed
Issue date: 2005-09-02
Publication year: 2005
ISBN: 91-7140-376-0
Statistics
Total Visits
Views | |
---|---|
Computational ...(legacy) | 329 |
Computational ... | 235 |
Total Visits Per Month
October 2023 | November 2023 | December 2023 | January 2024 | February 2024 | March 2024 | April 2024 | |
---|---|---|---|---|---|---|---|
Computational ... | 7 | 3 | 1 | 1 | 2 | 1 | 5 |
Top country views
Views | |
---|---|
United States | 101 |
Germany | 51 |
Sweden | 51 |
China | 48 |
United Kingdom | 14 |
Israel | 12 |
Netherlands | 12 |
India | 10 |
Japan | 10 |
South Korea | 10 |
Top cities views
Views | |
---|---|
Kiez | 15 |
Gothenburg | 11 |
Sunnyvale | 11 |
Beijing | 10 |
Seoul | 10 |
Shenzhen | 9 |
Tel Aviv | 7 |
Dublin | 6 |
Woodbridge | 6 |
Moscow | 5 |