Skip to main content

Thank you for visiting nature.com. 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.

  • Correspondence
  • Published:

Quantitative predictions of protein interactions with long noncoding RNAs

This is a preview of subscription content, access via your institution

Relevant articles

Open Access articles citing this article.

Access options

Rent or buy this article

Prices vary by article type

from$1.95

to$39.95

Prices may be subject to local taxes which are calculated during checkout

Figure 1: Training and testing of the Global Score for prediction of protein interactions with large RNAs.

Accession codes

Primary accessions

Gene Expression Omnibus

References

  1. Iyer, M.K. et al. Nat. Genet. 47, 199–208 (2015).

    Article  CAS  Google Scholar 

  2. Bellucci, M., Agostini, F., Masin, M. & Tartaglia, G.G. Nat. Methods 8, 444–445 (2011).

    Article  CAS  Google Scholar 

  3. Van Nostrand, E.L. et al. Nat. Methods 13, 508–514 (2016).

    Article  CAS  Google Scholar 

  4. Chu, C. et al. Cell 161, 404–416 (2015).

    Article  CAS  Google Scholar 

  5. McHugh, C.A. et al. Nature 521, 232–236 (2015).

    Article  CAS  Google Scholar 

  6. Minajigi, A. et al. Science 349, aab2276 (2015).

    Article  Google Scholar 

  7. Moindrot, B. et al. Cell Rep. 12, 562–572 (2015).

    Article  CAS  Google Scholar 

  8. Monfort, A. et al. Cell Rep. 12, 554–561 (2015).

    Article  CAS  Google Scholar 

Download references

Acknowledgements

We thank D. Whitworth and K. Havas for critical reading of the manuscript, the European Research Council (RIBOMYLOME_309545), MINECO (BFU2014-55054-P), Centro de Excelencia Severo Ochoa 2013–2017, and the European Molecular Biology Laboratory Grant-50800.

Author information

Authors and Affiliations

Authors

Corresponding authors

Correspondence to Andrea Cerase or Gian Gaetano Tartaglia.

Ethics declarations

Competing interests

The authors declare no competing financial interests.

Integrated supplementary information

Supplementary Figure 1 Global Scores performances on the test sets and comparison with other methods

A) Performances of computational methods (Area Under the ROC curve AUC) on proteins-RNAs interactions revealed by protein microarray technology. B) Xist interactions with RBPs reported by Minajigi et al., McHugh et al., Chu et al. (proteomic studies) as well as Moindrot et al. and Monfort et al. (genomic studies). For each set of protein and RNA fragments, we measured mean, median and maximum of the interaction propensities calculated with catRAPID and the binding score of RPIseq. Global Score outperforms catRAPID-based analyses and RPIseq for large lncRNAs (more details in Supplementary Tables 1,2,3,4).

Supplementary Figure 2 Comparison between predicted and eCLIP-validated interactions.

For 284 large transcripts (length >1000 nt), we studied the relationship between Global Score predictions and observed interactions revealed by eCLIP experiments in a) K562 and b) HepG2 cell lines. From low to high read counts, the fraction of interaction-prone RBPs (Global Score > 0.5) increases (upper plots; blue line) while RBPs with poor binding propensities (upper plots; red line; Global Score 0.5) show the opposite trend (log base 10 used for read counts; cubic function used for fitting). We assessed the significance of the trends by shuffling the read counts (bottom plots; black lines) and calculating two-sided Wilcoxon signed-rank test on predicted and randomized distributions. Global Score values are reported in Supplementary Table 5.

Supplementary Figure 3 Xist candidates selection.

We randomized the association between Global Score values and number of independent experiments reporting Xist interaction with a specific RBP (> 600 proteins used for the analysis; 10000 randomizations performed; see also Fig. 1d). Global Score values above 0.59 significantly discriminate 38 RBPs reported in at least two experimental assays (empirical p-value<0.01; Supplementary Table 6 and 7).

Supplementary Figure 4 Predictions of the RNA-binding domain of Lbr.

We ranked Lbr fragments by their interaction propensity to Xist lncRNA. The fragments corresponding to the top 10% of the statistical distribution are highlighted in yellow. The highest interaction propensity corresponds to amino acids 51-102 which corresponds to the RS domain implicated in nucleic acid recognition.

Supplementary Figure 5 RNA-binding regions of Spen, Hnrnpk, Hrnnpu/Saf-A, Lbr and Ptbp1.

Fragments overlapping with RNA-binding domains (RRM, KH, RGG and RS) rank high (top 2%) with respect to other protein regions (empirical p-values reported on the right). Fragments with the highest scores (top 2%) are coloured according to their interaction propensities.

Supplementary Figure 6 Predicted vs validated binding sites.

A) Relationship between Global Score values and areas under eCLIP profiles (Pearson correlation of 0.93 using the fitting formula QUOTE Global Score = α tanh (eCLIP) + β; p-value = 0.02). The areas are normalized relatively to the largest value of HnrnpK. B) Proximity of predicted binding sites to eCLIP peaks evaluated in terms of distance and overlap. Predicted fragments are in close proximity of eCLIP peaks and overlapping with them (significance of predictions is reported in Fig. 1e). The maximum distance observed (200 nt) is below the average distance between overlapping fragments (367 nt) and the maximum overlap corresponds to the fragment size (718 nt) used in our analysis.

Supplementary Figure 7 Significance of Global Score predictions.

We compared interaction propensities of target candidates with a large set of nucleotide-binding proteins. From low to high Global Score values, the ratio of identified candidates over number of predicted interactions increases monotonically, reaching 50% at the 99th Global Score percentile (p-value = 10 −8) and 100% at the 99.9th percentile (p-value = 10−20; Supplementary Table 9).

Supplementary information

Supplementary Text and Figures

Supplementary Figures 1–7, and Supplementary Table 4, 6, 8 and 9 (PDF 7339 kb)

Supplementary Table 1

Protein and RNA sequences used for training, test and Xist sets. (XLSX 1580 kb)

Supplementary Table 2

Global Score values for the training, test and Xist sets. (XLSX 170 kb)

Supplementary Table 3

Performances of computational methods on the test set and Xist interactions. (XLSX 40 kb)

Supplementary Table 5

Global Score predictions of RBP interactions with large transcripts (XLSX 131 kb)

Supplementary Table 7

Predicted interactions in proteomic and genetic screenings. (XLSX 45 kb)

Source data

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Cirillo, D., Blanco, M., Armaos, A. et al. Quantitative predictions of protein interactions with long noncoding RNAs. Nat Methods 14, 5–6 (2017). https://doi.org/10.1038/nmeth.4100

Download citation

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1038/nmeth.4100

This article is cited by

Search

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