Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2012 Dec;40(22):11673-83.
doi: 10.1093/nar/gks901. Epub 2012 Oct 2.

MiRmap: comprehensive prediction of microRNA target repression strength

Charles E Vejnar  1 , Evgeny M Zdobnov
Affiliations

MiRmap: comprehensive prediction of microRNA target repression strength

Charles E Vejnar et al. Nucleic Acids Res. 2012 Dec.

Abstract

MicroRNAs, or miRNAs, post-transcriptionally repress the expression of protein-coding genes. The human genome encodes over 1000 miRNA genes that collectively target the majority of messenger RNAs (mRNAs). Base pairing of the so-called miRNA 'seed' region with mRNAs identifies many thousands of putative targets. Evaluating the strength of the resulting mRNA repression remains challenging, but is essential for a biologically informative ranking of potential miRNA targets. To address these challenges, predictors may use thermodynamic, evolutionary, probabilistic or sequence-based features. We developed an open-source software library, miRmap, which for the first time comprehensively covers all four approaches using 11 predictor features, 3 of which are novel. This allowed us to examine feature correlations and to compare their predictive power in an unbiased way using high-throughput experimental data from immunopurification, transcriptomics, proteomics and polysome fractionation experiments. Overall, target site accessibility appears to be the most predictive feature. Our novel feature based on PhyloP, which evaluates the significance of negative selection, is the best performing predictor in the evolutionary category. We combined all the features into an integrated model that almost doubles the predictive power of TargetScan. miRmap is freely available from http://cegg.unige.ch/mirmap.

PubMed Disclaimer

Figures

Figure 1.
Figure 1.
miRmap library usage: after importing the library (lines 1 and 2), a ‘mimset’ object is created containing the mRNA and miRNA sequences. We then call a method of the mimset object to search (line 5) for seeds with a length of 7 (all parameters have defaults that can be changed this way). The link with the C libraries is initalized on line 7. We then manually evaluate the repression strength with differents methods (lines 9–16). Each of these methods have modifiable parameters. We finally print a report (line 18).
Figure 2.
Figure 2.
Correlation among features based on prediction for human miRNAs and mRNAs. (A) A heatmap of the absolute values of Spearman correlation coefficients between pairs of features classified in methods categories. Venn diagrams (B) and (C) show the overlaps among the first best prediction quartiles of selected features. One feature per category (sequence-based with ‘AU content’, conservation with the ‘BLS’ and probabilistic with ‘P.over exact’) is shown on (A). Venn diagram (C) underlines the high overlap between ‘AU content’ and ‘ΔG open’ that we grouped in the ‘accessibility group’, whereas ‘ΔG duplex’ has a very low overlap with these two features. We grouped ‘ΔG duplex’ with ‘ΔG binding’ in the ‘binding energy’ group. Numbers of predicted relationships between human miRNA and mRNA are written in the corresponding overlaps of the Venn diagrams.
Figure 3.
Figure 3.
Correlation between each feature and the expression fold-changes of mRNAs following miRNA injection (‘Trans.Grimson’ dataset). Data points were binned in 15 equally sized bins. The average in each bin is represented by a blue dot. We fitted a linear regression model (red line) on the blue dots. r is the correlation on the full dataset; r′ is the correlation on the binned dataset. P-values can be found in Supplementary Table S2.
Figure 4.
Figure 4.
Correlation between each feature and the seven experimental miRNA repression measures (the name of the first author of each dataset is shown in grey) classified in transcriptomics, proteomics, IP and polysome fractionation experiment types. Target prediction features are organized into groups that aim to evaluate the same type of information. The radial axis represents the correlation coefficient (the highest correlations are the furthest from the centre of the circle).
Figure 5.
Figure 5.
(A) Performance comparison (as coefficient correlations with experimental miRNA repression measures; order of the experiments is the same as Figure 4) of the best performing feature (brown), TargetScan context score (red) and miRmap (blue). (B) Feature relative importance in the miRmap multiple linear regression model predicting miRNA repression strength. R2 is the proportion of variance explained by the model. ‘AU content’ is the most explanatory variable with 29% of R2.
See this image and copyright information in PMC

Similar articles

See all similar articles

Cited by

See all "Cited by" articles

References

    1. Bartel DP. MicroRNAs: target recognition and regulatory functions. Cell. 2009;136:215–233. - PMC - PubMed
    1. Muljo SA, Kanellopoulou C, Aravind L. MicroRNA targeting in mammalian genomes: genes and mechanisms. Wiley Interdiscip. Rev. Syst. Biol. Med. 2010;2:148–161. - PMC - PubMed
    1. Axtell MJ, Westholm JO, Lai EC. Vive la différence: biogenesis and evolution of microRNAs in plants and animals. Genome Biol. 2011;12:221. - PMC - PubMed
    1. Brennecke J, Stark A, Russell RB, Cohen SM. Principles of microRNA-target recognition. PLoS Biol. 2005;3:e85. - PMC - PubMed
    1. Didiano D, Hobert O. Perfect seed pairing is not a generally reliable predictor for miRNA-target interactions. Nat. Struct. Mol. Biol. 2006;13:849–851. - PubMed

Publication types