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.

Prediction of drug combination effects with a minimal set of experiments


High-throughput drug combination screening provides a systematic strategy to discover unexpected combinatorial synergies in pre-clinical cell models. However, phenotypic combinatorial screening with multi-dose matrix assays is experimentally expensive, especially when the aim is to identify selective combination synergies across a large panel of cell lines or patient samples. Here, we implement DECREASE, an efficient machine learning model that requires only a limited set of pairwise dose–response measurements for accurate prediction of drug combination synergy in a given sample. Using a compendium of 23,595 drug combination matrices tested in various cancer cell lines and malaria and Ebola infection models, we demonstrate how cost-effective experimental designs with DECREASE capture almost the same degree of information for synergy and antagonism detection as the fully measured dose–response matrices. Measuring only the matrix diagonal provides an accurate and practical option for combinatorial screening. The minimal-input web implementation enables applications of DECREASE to both pre-clinical and translational studies.

Access options

Rent or Buy article

Get time limited or full article access on ReadCube.


All prices are NET prices.

Fig. 1: Schematic representation of the computational steps implemented in the DECREASE approach.
Fig. 2: DECREASE accurately predicts combinatorial responses with cost-effective experimental HTS designs.
Fig. 3: DECREASE accurately predicts drug combination landscapes with a fixed-concentration design.
Fig. 4: Selection and use of dose–response matrix rows for the prediction of drug combination effects.
Fig. 5: DECREASE accurately predicts combinatorial responses in antimalaria and antiviral applications.

Data availability

The unpublished in-house anticancer dose–response matrix data for all 210 combinations are available either from the DECREASE web tool ( or the GitHub repository ( The published DLBCL anticancer dose–response matrix data are available at the Tripod portal ( The data for 22,737 anticancer drug combination from the study of O’Neil and others3 were downloaded from the original paper ( The published antimalarial dose–response matrix data are available at the Tripod repository ( The published antiviral dose–response matrix data for Ebola treatment are available from the NCATS Matrix portal (

Code availability

The source code of the DECREASE prediction algorithm is freely available either at the tool website ( or GitHub ( to allow replication of the results and to compare or combine the cNMF algorithm with other prediction models. The source codes of the other algorithms are publicly available (web links and package versions are listed in Supplementary Table 3). The recent Dose model33 implementation was provided by request from its authors, A. Zimmer and U. Alon.


  1. 1.

    Webster, R. M. Combination therapies in oncology. Nat. Rev. Drug Discov. 15, 81–82 (2016).

    Google Scholar 

  2. 2.

    Lehár, J. et al. Synergistic drug combinations tend to improve therapeutically relevant selectivity. Nat. Biotechnol. 27, 659–666 (2009).

    Google Scholar 

  3. 3.

    O’Neil, J. et al. An unbiased oncology compound screen to identify novel combination strategies. Mol. Cancer Ther. 15, 1155–1162 (2016).

    Google Scholar 

  4. 4.

    Radic-Sarikas, B. et al. Combinatorial drug screening identifies Ewing sarcoma-specific sensitivities. Mol. Cancer Ther. 16, 88–101 (2017).

    Google Scholar 

  5. 5.

    Pelaia, G., Vatrella, A. & Maselli, R. The potential of biologics for the treatment of asthma. Nat. Rev. Drug Discov. 11, 958–972 (2012).

    Google Scholar 

  6. 6.

    Chakradhar, S. All in one: researchers create combination drugs for diabetes and obesity. Nat. Med. 22, 694–696 (2016).

    Google Scholar 

  7. 7.

    Worthington, R. J. & Melander, C. Combination approaches to combat multidrug-resistant bacteria. Trends Biotechnol. 31, 177–184 (2013).

    Google Scholar 

  8. 8.

    Chandrasekaran, S. et al. Chemogenomics and orthology-based design of antibiotic combination therapies. Mol. Syst. Biol. 12, 872 (2016).

    Google Scholar 

  9. 9.

    Borisy, A. A. et al. Systematic discovery of multicomponent therapeutics. Proc. Natl Acad. Sci. USA 100, 7977–7982 (2003).

    Google Scholar 

  10. 10.

    Mott, B. T. et al. High-throughput matrix screening identifies synergistic and antagonistic antimalarial drug combinations. Sci. Rep. 5, 13891 (2015).

    Google Scholar 

  11. 11.

    Mathews Griner, L. A. et al. High-throughput combinatorial screening identifies drugs that cooperate with ibrutinib to kill activated B-cell-like diffuse large B-cell lymphoma cells. Proc. Natl Acad. Sci. USA 111, 2349–2354 (2014).

    Google Scholar 

  12. 12.

    Janzen, W. P. Screening technologies for small molecule discovery: the state of the art. Chem. Biol. 21, 1162–1170 (2014).

    Google Scholar 

  13. 13.

    He, L. et al. Patient-customized drug combination prediction and testing for T-cell prolymphocytic leukemia patients. Cancer Res. 78, 2407–2418 (2018).

    Google Scholar 

  14. 14.

    Holbeck, S. L. et al. The national cancer institute ALMANAC: a comprehensive screening resource for the detection of anticancer drug pairs with enhanced therapeutic activity. Cancer Res. 13, 3564–3576 (2017).

    Google Scholar 

  15. 15.

    Roy, A. et al. Open access high throughput drug discovery in the public domain: a Mount Everest in the making. Curr. Pharm. Biotechnol. 11, 764–778 (2010).

    Google Scholar 

  16. 16.

    Kurtz, S. E. et al. Molecularly targeted drug combinations demonstrate selective effectiveness for myeloid- and lymphoid-derived hematologic malignancies. Proc. Natl Acad. Sci. USA 114, 7554–7563 (2017).

    Google Scholar 

  17. 17.

    Yu, D. et al. Identification of synergistic, clinically achievable, combination therapies for osteosarcoma. Sci. Rep. 5, 16991 (2015).

    Google Scholar 

  18. 18.

    Horn, T. et al. High-order drug combinations are required to effectively kill colorectal cancer cells. Cancer Res. 76, 6950–6963 (2016).

    Google Scholar 

  19. 19.

    Chou, A. et al. Tailored first-line and second-line CDK4-targeting treatment combinations in mouse models of pancreatic cancer. Gut 67, 2142–2155 (2018).

    Google Scholar 

  20. 20.

    Chou, Ting-Chao Theoretical basis, experimental design, and computerized simulation of synergism and antagonism in drug combination studies. Pharmacol. Rev. 58, 621–681 (2006).

    Google Scholar 

  21. 21.

    Sun, W. et al. Rapid antimicrobial susceptibility test for identification of new therapeutics and drug combinations against multidrug-resistant bacteria. Emerg. Microbes Infect. 5, e116 (2016).

    Google Scholar 

  22. 22.

    Shehata, M. et al. Reconstitution of PTEN activity by CK2 inhibitors and interference with the PI3-K/Akt cascade counteract the antiapoptotic effect of human stromal cells in chronic lymphocytic leukemia. Blood 116, 2513–2521 (2010).

    Google Scholar 

  23. 23.

    Tan, X. et al. Systematic identification of synergistic drug pairs targeting HIV. Nat. Biotechnol. 30, 1125–1130 (2012).

    Google Scholar 

  24. 24.

    Di Veroli, G. Y. et al. Combenefit: an interactive platform for the analysis and visualization of drug combinations. Bioinformatics 32, 2866–2868 (2016).

    Google Scholar 

  25. 25.

    Ianevski, A., He, L., Aittokallio, T. & Tang, J. SynergyFinder: a web application for analyzing drug combination dose–response matrix data. Bioinformatics 33, 2413–2415 (2017).

    Google Scholar 

  26. 26.

    Loewe, S. The problem of synergism and antagonism of combined drugs. Arzneimiettelforschung 3, 286–290 (1953).

    Google Scholar 

  27. 27.

    Bliss, C. I. The toxicity of poisons applied jointly. Ann. Appl. Biol. 26, 585–615 (1939).

    Google Scholar 

  28. 28.

    Berenbaum, M. C. What is synergy? Pharmacol. Rev. 41, 93–141 (1989).

    Google Scholar 

  29. 29.

    Yadav, B. et al. Searching for drug synergy in complex dose-response landscapes using an interaction potency model. Comput. Struct. Biotechnol. J. 13, 504–505 (2015).

    Google Scholar 

  30. 30.

    Chevereau, G. & Bollenbach, T. Systematic discovery of drug interaction mechanisms. Mol. Syst. Biol. 11, 807 (2015).

    Google Scholar 

  31. 31.

    Chou, T. C. Drug combination studies and their synergy quantification using the Chou–Talalay method. Cancer Res. 70, 440–446 (2010).

    Google Scholar 

  32. 32.

    Al-Lazikani, B. et al. Combinatorial drug therapy for cancer in the post-genomic era. Nat. Biotechnol. 30, 679–692 (2012).

    Google Scholar 

  33. 33.

    Zimmer, A. et al. Prediction of multidimensional drug dose responses based on measurements of drug pairs. Proc. Natl Acad. Sci. USA 113, 10442–10447 (2016).

    Google Scholar 

  34. 34.

    Eziefula, A. C. Artesunate-mefloquine: a malaria treatment for African children? Lancet Infect. Dis. 16, 1086–1087 (2016).

    Google Scholar 

  35. 35.

    Dyall, J. et al. Identification of combinations of approved drugs with synergistic activity against Ebola virus in cell cultures. J. Infect. Dis. 218, S672–S678 (2018).

    Google Scholar 

  36. 36.

    Weinstein, Z. B. et al. Prediction of synergistic drug combinations. Curr. Opin. Syst. Biol. 4, 24–28 (2015).

    Google Scholar 

  37. 37.

    Szwajda, A. et al. Systematic mapping of kinase addiction combinations in breast cancer cells by integrating drug sensitivity and selectivity profiles. Chem. Biol. 22, 1144–1155 (2015).

    Google Scholar 

  38. 38.

    Mullard, A. Microfluidics platform lowers barrier to drug combination screening. Nat. Rev. Drug Discov. 17, 691–692 (2018).

    Google Scholar 

  39. 39.

    Gautam, P. et al. Identification of selective cytotoxic and synthetic lethal drug responses in triple negative breast cancer cells. Mol. Cancer 15, 34 (2016).

    MathSciNet  Google Scholar 

  40. 40.

    Gadagkar, S. R. & Call, G. B. Computational tools for fitting the Hill equation to dose–response curves. J. Pharmacol. Toxicol. Methods 71, 68–76 (2015).

    Google Scholar 

  41. 41.

    Hill, A. V. The possible effects of the aggregation of the molecules of hemoglobin on its dissociation curves. J. Physiol. 40, 4–7 (1910).

    Google Scholar 

  42. 42.

    Ritz, C., Baty, F., Streibig, J. C. & Gerhard, D. Dose–response analysis using R. PLoS One 10, e0146021 (2015).

    Google Scholar 

  43. 43.

    Wang, Y. X. et al. Non-negative matrix factorization: a comprehensive review. IEEE Trans. Knowl. Data Eng. 25, 1336–1353 (2013).

    Google Scholar 

  44. 44.

    Venter, J. H. On estimation of the mode. Ann. Math. Stat. 38, 1446–1455 (1967).

    MathSciNet  MATH  Google Scholar 

  45. 45.

    Tang, J., Wennerberg, K. & Aittokallio, T. What is synergy? The Saariselkä agreement revisited. Front. Pharmacol. 6, 181 (2015).

    Google Scholar 

  46. 46.

    Bischl, B. et al. mlrMBO: a modular framework for model-based optimization of expensive black-box functions. Preprint at (2017).

  47. 47.

    Chen, T. & Guestrin, C. Xgboost: a scalable tree boosting system. In Proc. 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining 785–794 (ACM, 2016).

Download references


We thank the authors of the original publications (refs. 3,10,11,35) for making their drug combination data matrices publicly available. We thank the authors of the Dose model publication (ref. 33) for providing their MATLAB implementation and guaranteeing its appropriate use in this work, D. Bulanova (FIMM) for her valuable comments and discussions and O. Hansson (FIMM) for technical support with the web server. This work was funded by the Academy of Finland (grants nos. 272577 and 277293 to K.W. and 292611, 279163, 295504, 310507, 313267 and 326238 to T.A.), the European Union’s Horizon 2020 Research and Innovation Programme (ERA PerMed JAKSTAT-TARGET), the Cancer Society of Finland (T.A. and K.W.), Sigrid Jusélius Foundation (K.W. and T.A.) and Novo Nordisk Foundation (NNF17CC0027852 to K.W.). The FIMM High Throughput Biomedine Unit is supported financially by the University of Helsinki and Biocenter Finland (S.P. and J.S.).

Author information




T.A., K.W. and A.I. designed and conceived the study. A.I. developed cNMF and implemented the web platform. A.I. and A.K.G. selected and tested various machine learning methods. P.G. performed the drug combination screening experiments. A.I., A.K.G. and A.K. managed data integration from various sources. A.I. and A.K. prepared the figures and tables for the manuscript. A.I., A.K.G. and T.A. wrote the manuscript. A.K., P.G., S.P., J.S. and K.W. critically reviewed and edited the manuscript. All authors reviewed and approved the final version of the manuscript. A.K.G. and P.G. contributed equally.

Corresponding author

Correspondence to Tero Aittokallio.

Ethics declarations

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary information

Supplementary Information

Supplementary figures, tables and methods

Reporting Summary

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Ianevski, A., Giri, A.K., Gautam, P. et al. Prediction of drug combination effects with a minimal set of experiments. Nat Mach Intell 1, 568–577 (2019).

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