Skip to main content

Thank you for visiting You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.

Design of a genome-wide siRNA library using an artificial neural network

A Corrigendum to this article was published on 01 August 2006

A Corrigendum to this article was published on 01 October 2005


The largest gene knock-down experiments performed to date have used multiple short interfering/short hairpin (si/sh)RNAs per gene1,2,3. To overcome this burden for design of a genome-wide siRNA library, we used the Stuttgart Neural Net Simulator to train algorithms on a data set of 2,182 randomly selected siRNAs targeted to 34 mRNA species, assayed through a high-throughput fluorescent reporter gene system. The algorithm, (BIOPREDsi), reliably predicted activity of 249 siRNAs of an independent test set (Pearson coefficient r = 0.66) and siRNAs targeting endogenous genes at mRNA and protein levels. Neural networks trained on a complementary 21-nucleotide (nt) guide sequence were superior to those trained on a 19-nt sequence. BIOPREDsi was used in the design of a genome-wide siRNA collection with two potent siRNAs per gene. When this collection of 50,000 siRNAs was used to identify genes involved in the cellular response to hypoxia, two of the most potent hits were the key hypoxia transcription factors HIF1A and ARNT.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Figure 1: Thirty seven siRNAs were designed and tested against three human targets (for siRNA sequences, see Supplementary Table 1).
Figure 2: Management of data during the training and testing of artificial neural networks (ANN).
Figure 3: Performance of the algorithm with respect to both ranking 36 siRNAs of various potencies and selecting potent siRNAs.


  1. 1

    Paddison, P. et al. A resource for large-scale RNA-interference-based screens in mammals. Nature 428, 427–431 (2004).

    CAS  Article  Google Scholar 

  2. 2

    Berns, K. et al. A large-scale RNAi screen in human cells identifies new components of the p53 pathway. Nature 428, 431–437 (2004).

    CAS  Article  Google Scholar 

  3. 3

    Kittler, R. et al. An endoribonuclease-prepared siRNA screen in human cells identifies genes essential for cell division. Nature 432, 1036–1040 (2004).

    CAS  Article  Google Scholar 

  4. 4

    Boese, B. et al. Mechanistic insights aid computational short interfering RNA design. Methods Enzymol. 392, 73–96 (2005).

    CAS  Article  Google Scholar 

  5. 5

    Khvorova, A., Reynolds, A. & Jayasena, S.D. Functional siRNAs and miRNAs exhibit strand bias. Cell 115, 209–216 (2003).

    CAS  Article  Google Scholar 

  6. 6

    Schwarz, D.S. et al. Asymmetry in the assembly of the RNAi enzyme complex. Cell 115, 199–208 (2003).

    CAS  Article  Google Scholar 

  7. 7

    Reynolds, A. et al. Rational siRNA design for RNA interference. Nat. Biotechnol. 22, 326–330 (2004).

    CAS  Article  Google Scholar 

  8. 8

    Hsieh, A.C. et al. A library of siRNA duplexes targeting the phosphoinositide 3-kinase pathway: determinants of gene silencing for use in cell-based screens. Nucleic Acids Res. 32, 893–901 (2004).

    CAS  Article  Google Scholar 

  9. 9

    Ui-Tei, K. et al. Guidelines for the selection of highly effective siRNA sequences for mammalian and chick RNA interference. Nucleic Acids Res. 32, 936–948 (2004).

    CAS  Article  Google Scholar 

  10. 10

    Amarzguioui, M. & Prydz, H. An algorithm for selection of functional siRNA sequences. Biochem. Biophys. Res. Commun. 316, 1050–1058 (2004).

    CAS  Article  Google Scholar 

  11. 11

    Saetrom, P. & Snove, O., Jr. A comparison of siRNA efficacy predictors. Biochem. Biophys. Res. Commun. 321, 247–253 (2004).

    CAS  Article  Google Scholar 

  12. 12

    Labuda, D., Nicoghosian, K. & Cedergren, R.J. A novel RNA digesting activity from commercial polynucleotide phosphorylase. FEBS Lett. 179, 213–216 (1985).

    CAS  Article  Google Scholar 

  13. 13

    Kierzek, R. Hydrolysis of oligoribonucleotides: influence of sequence and length. Nucleic Acids Res. 20, 5073–5077 (1992).

    CAS  Article  Google Scholar 

  14. 14

    Schneider, G. & Wrede, P. Artificial neural networks for computer-based molecular design. Prog. Biophys. Mol. Biol. 70, 175–222 (1998).

    CAS  Article  Google Scholar 

  15. 15

    Weinstein, J.N. et al. Neural computing in cancer drug development: predicting mechanism of action. Science 258, 447–451 (1992).

    CAS  Article  Google Scholar 

  16. 16

    Stolorz, P., Lapedes, A. & Xia, Y. Predicting protein secondary structure using neural net and statistical methods. J. Mol. Biol. 225, 363–377 (1992).

    CAS  Article  Google Scholar 

  17. 17

    Giddings, M.C. et al. Artificial neural network prediction of antisense oligodeoxynucleotide activity. Nucleic Acids Res. 30, 4295–4304 (2002).

    CAS  Article  Google Scholar 

  18. 18

    Chalk, A.M. & Sonnhammer, E.L. Computational antisense oligo prediction with a neural network model. Bioinformatics 18, 1567–1575 (2002).

    CAS  Article  Google Scholar 

  19. 19

    Rumelhart, D. in Parallel distributed processing (eds. McClelland, J. & the PDP Research Group), vol. 1, 318–362, (MIT Press, Cambridge, MA).

  20. 20

    Huesken, D. et al. mRNA fusion constructs serve in a general cell-based assay to profile oligonucleotide activity. Nucleic Acids Res. 31, e102/1–e102/11 (2003).

    CAS  Google Scholar 

  21. 21

    Zamore, P.D., Sharp, P.A., Tuschl, T. & Bartel, D.P. RNAi: double-stranded RNA directs the ATP-dependent cleavage of mRNA at 21 to 23 nucleotide intervals. Cell 101, 25–33 (2000).

    CAS  Article  Google Scholar 

  22. 22

    Benjamini, Y. & Hochberg, Y. Controlling the false discovery rate: a practical and powerful approach to multiple testing. J. of the Royal Statistical Society. Series B 57, 289–300 (1995).

    Google Scholar 

  23. 23

    Vickers, T.A. et al. Efficient reduction of target RNAs by small interfering RNA and RNASE H-dependent antisense agents. J. Biol. Chem. 278, 7108–7118 (2003).

    CAS  Article  Google Scholar 

  24. 24

    Horbarth, J. et al. Sequence, chemical and structural variation of small interfering RNAs and short hairpin RNAs and the effect on mammalian gene silencing. Antisense Nucleic Acid Drug Dev. 13, 83–106 (2003).

    Article  Google Scholar 

  25. 25

    Ohba, H. et al. Inhibition of bcr-abl and/or c-abl gene expression by small interfering, double-stranded RNAs: cross-talk with cell proliferation factors and other oncogenes. Cancer 101, 1390–1403 (2004).

    CAS  Article  Google Scholar 

  26. 26

    Hall, J. Unraveling the general properties of siRNAs: strength in numbers and lessons from the past. Nat. Rev. Genet. 5, 552–557 (2004).

    CAS  Article  Google Scholar 

  27. 27

    Hemmings-Mieszczak, M., Dorn, G., Natt, F., Hall, J. & Wishart, W. Antisense oligonucleotides complement RNAi-mediated specific inhibition of the recombinant rat P2X3 receptor. Nucleic Acids Res. 31, 2117–2126 (2003).

    CAS  Article  Google Scholar 

  28. 28

    Butz, N. et al. The human ubiquitin-conjugating enzyme Cdc34 controls cellular proliferation through regulation of p27Kip1 protein levels. Exp. Cell Res. 303, 482–493 (2005).

    CAS  Article  Google Scholar 

Download references


We thank Erik Bury for preliminary work on development of the algorithm. We also thank B. Csordas, V. Drephal, A. Garnier, S. Gfeller, D. Kirk, B. Pak, R. Theurillat, R. Widmer, S. Zhao and W. Zuercher for excellent technical support. We thank J. Mestan and F. Hofmann for E2 antibodies. We thank P. Weiss for helpful discussions and J. Hunziker and R. Haener for carefully reading the manuscript.

Author information



Corresponding author

Correspondence to Jonathan Hall.

Ethics declarations

Competing interests

D.H., J.L., C.M., J.W., F.A., J.W., M.L., M.R., D.C., F.N. and J.H. are employees of Novartis. A.R., S.E. and B.M. are current or former employees of Compugen.

Supplementary information

Supplementary Fig. 1

Dual reporter assay used to generate a data set for training of neural networks. (PDF 473 kb)

Supplementary Table 1

siRNAs targeted to endogenous genes and used in experiments displayed in figures 1a-i and figure 3a-c. (PDF 48 kb)

Supplementary Table 2

cDNA insert sequences used for siRNA design. (PDF 131 kb)

Supplementary Table 3

siRNA sequences and their normalized inhibitory activities used for training or testing of ANNs. (PDF 618 kb)

Supplementary Table 4

Single nucleotide (nt) motifs overrepresented with significance lower than 5% in the 200 most potent and 200 least potent siRNAs were taken from Supplementary Table 3. (PDF 56 kb)

Supplementary Table 5

siRNAs targeted to endogenous genes and used in experiments displayed in figures 3d-k. (PDF 51 kb)

Supplementary Table 6

Q-PCR primer sequences. (PDF 41 kb)

Supplementary Table 7

ABI Assays-on-demand. (PDF 43 kb)

Rights and permissions

Reprints and Permissions

About this article

Cite this article

Huesken, D., Lange, J., Mickanin, C. et al. Design of a genome-wide siRNA library using an artificial neural network. Nat Biotechnol 23, 995–1001 (2005).

Download citation

Further reading


Quick links

Nature Briefing

Sign up for the Nature Briefing newsletter — what matters in science, free to your inbox daily.

Get the most important science stories of the day, free in your inbox. Sign up for Nature Briefing