Skip to main page content
Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2017 Aug 21;45(14):8541-8550.
doi: 10.1093/nar/gkx512.

Advanced Multi-Loop Algorithms for RNA Secondary Structure Prediction Reveal That the Simplest Model Is Best

Affiliations
Free PMC article

Advanced Multi-Loop Algorithms for RNA Secondary Structure Prediction Reveal That the Simplest Model Is Best

Max Ward et al. Nucleic Acids Res. .
Free PMC article

Abstract

Algorithmic prediction of RNA secondary structure has been an area of active inquiry since the 1970s. Despite many innovations since then, our best techniques are not yet perfect. The workhorses of the RNA secondary structure prediction engine are recursions first described by Zuker and Stiegler in 1981. These have well understood caveats; a notable flaw is the ad-hoc treatment of multi-loops, also called helical-junctions, that persists today. While several advanced models for multi-loops have been proposed, it seems to have been assumed that incorporating them into the recursions would lead to intractability, and so no algorithms for these models exist. Some of these models include the classical model based on Jacobson-Stockmayer polymer theory, and another by Aalberts and Nadagopal that incorporates two-length-scale polymer physics. We have realized practical, tractable algorithms for each of these models. However, after implementing these algorithms, we found that no advanced model was better than the original, ad-hoc model used for multi-loops. While this is unexpected, it supports the praxis of the current model.

Figures

Figure 1.
Figure 1.
A tRNA secondary structure (Sprinzl ID RA1661 (43)) that is perfectly predicted by the linear model (PPV = 1, sensitivity = 1), but not by the logarithmic model (PPV = 0, sensitivity = 0). Panel (A) shows the prediction from the linear model and panel (B) shows the prediction from the logarithmic model. The logarithmic model computes the free energy of the linear prediction to be –30.1 kcal/mol, while its own prediction has a score of –30.2 kcal/mol. The linear model, on the other hand gives these scores of –30.9 kcal/mol and –30.2 kcal/mol respectively.
Figure 2.
Figure 2.
An SRP RNA structure that is almost perfectly predicted by the linear model (PPV = 0.929, sensitivity = 0.963), and poorly predicted by the AN model (PPV = 0, sensitivity = 0). Panel (A) is the accepted structure, panel (B) is the structure predicted by the linear model and panel (C) is the structure predicted by the AN model (35). The AN model gives the linear prediction a score of –32.9 kcal/mol, while its own prediction gets a score of –33.4 kcal/mol. The linear model gives scores of –32.9 kcal/mol and –32.3 kcal/mol respectively.

Similar articles

See all similar articles

Cited by 2 articles

References

    1. Eddy S.R. Computational genomics of noncoding RNA genes. Cell. 2002; 109:137–140. - PubMed
    1. Hofacker I., Stadler P.F. RNAz 2.0: improved noncoding RNA detection. Pacific Symposium on Biocomputing. 2010; 15:69–79. - PubMed
    1. Fu Y., Xu Z.Z., Lu Z.J., Zhao S., Mathews D.H. Discovery of novel ncRNA sequences in multiple genome alignments on the basis of conserved and stable secondary structures. PLoS One. 2015; 10:e0130200. - PMC - PubMed
    1. Kodama Y., Shumway M., Leinonen R. The sequence read archive: explosive growth of sequencing data. Nucleic Acids Res. 2012; 40:D54–D56. - PMC - PubMed
    1. Pace N.R., Thomas B.C., Woese C.R. The RNA World. 1999; 2nd edn, NY: Cold Spring Harbor Laboratory Press.
Feedback