Limits...
A folding algorithm for extended RNA secondary structures.

Hรถner zu Siederdissen C, Bernhart SH, Stadler PF, Hofacker IL - Bioinformatics (2011)

Bottom Line: Successful prediction of these structural features leads to improved secondary structures with applications in tertiary structure prediction and simultaneous folding and alignment.We accompany this model with a number of programs for parameter optimization and structure prediction.All sources (optimization routines, RNA folding, RNA evaluation, extended secondary structure visualization) are published under the GPLv3 and available at www.tbi.univie.ac.at/software/rnawolf/.

View Article: PubMed Central - PubMed

Affiliation: Institute for Theoretical Chemistry, University of Vienna, A-1090 Vienna, Austria. choener@tbi.inivie.ac.at

ABSTRACT

Motivation: RNA secondary structure contains many non-canonical base pairs of different pair families. Successful prediction of these structural features leads to improved secondary structures with applications in tertiary structure prediction and simultaneous folding and alignment.

Results: We present a theoretical model capturing both RNA pair families and extended secondary structure motifs with shared nucleotides using 2-diagrams. We accompany this model with a number of programs for parameter optimization and structure prediction.

Availability: All sources (optimization routines, RNA folding, RNA evaluation, extended secondary structure visualization) are published under the GPLv3 and available at www.tbi.univie.ac.at/software/rnawolf/.

Show MeSH
MC-Fold and MC-Fold-DP both consider small loops, like the hairpin AAGUG (๐’ž) and the 2ร—2 stack AAGU (๐’Ÿ) (read clockwise, starting bottom left). Each loop is scored by a function Ec(๐’ž/AAGUG). The stack (๐’Ÿ) follows analoguously. The interaction term between two loops is calculated as indicated by the arrow (ฮฑ), where the two loops are overlayed at the common AG pair. The contribution of the interaction is Ejunction+hinge(๐’ž,๐’Ÿ;ฮธ;A,G) with ฮธ the unknown pair family.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3117358&req=5

Figure 2: MC-Fold and MC-Fold-DP both consider small loops, like the hairpin AAGUG (๐’ž) and the 2ร—2 stack AAGU (๐’Ÿ) (read clockwise, starting bottom left). Each loop is scored by a function Ec(๐’ž/AAGUG). The stack (๐’Ÿ) follows analoguously. The interaction term between two loops is calculated as indicated by the arrow (ฮฑ), where the two loops are overlayed at the common AG pair. The contribution of the interaction is Ejunction+hinge(๐’ž,๐’Ÿ;ฮธ;A,G) with ฮธ the unknown pair family.

Mentions: Like ordinary secondary structure prediction tools, MC-Fold (Parisien and Major, 2008) is based on a decomposition of the RNA structure into โ€˜loopsโ€™. In contrast to the standard energy model, however, it considers the full set of base pair types available in the LW representation. Each base pair, therefore, corresponds to a triple (i,j;ฮธ) where ฮธ is one of the 12 types of pairs. In this model, ordinary secondary structures are the subset of pairs with Watson-Crick type (ฮธ=โ€˜WWโ€™) and the two nucleotides form one of the six canonical combinations {AU,UA,CG,GC,GU,UG{. This extension of the structure model also calls for a more sophisticated energy model. While the standard model assumes the contributions of the loops to be strictly additive, MC-Fold also considers interactions between adjacent unbranched loops (hairpins, stacked pairs, bulges and general interior loops). This means that the total energy of a structure is not only dependent on the loop types present, but also on the arrangement of these loops. Dispensing with details of the parametrization, the scoring function of MC-Fold for a structure ๐’ฎ on sequence x can be written as follows (see Fig. 2):(2.1)where ๐’ž,๐’žโ€ฒ,๐’žโ€ฒ are different loops of ๐’ฎ. The additive term Ec tabulates the (sequence-dependent) contributions of the loops. The interaction term Ej+h accounts for the โ€˜junctionโ€™ and โ€˜hingeโ€™ terms in stemโ€“loop regions. These interaction terms depend on the type of the adjacent loops as well as on the type ฮธ and sequence (x[k], x[l]) of the base pair that connects them. For multiloops, only the additive term is considered.Fig. 2.


A folding algorithm for extended RNA secondary structures.

Hรถner zu Siederdissen C, Bernhart SH, Stadler PF, Hofacker IL - Bioinformatics (2011)

MC-Fold and MC-Fold-DP both consider small loops, like the hairpin AAGUG (๐’ž) and the 2ร—2 stack AAGU (๐’Ÿ) (read clockwise, starting bottom left). Each loop is scored by a function Ec(๐’ž/AAGUG). The stack (๐’Ÿ) follows analoguously. The interaction term between two loops is calculated as indicated by the arrow (ฮฑ), where the two loops are overlayed at the common AG pair. The contribution of the interaction is Ejunction+hinge(๐’ž,๐’Ÿ;ฮธ;A,G) with ฮธ the unknown pair family.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3117358&req=5

Figure 2: MC-Fold and MC-Fold-DP both consider small loops, like the hairpin AAGUG (๐’ž) and the 2ร—2 stack AAGU (๐’Ÿ) (read clockwise, starting bottom left). Each loop is scored by a function Ec(๐’ž/AAGUG). The stack (๐’Ÿ) follows analoguously. The interaction term between two loops is calculated as indicated by the arrow (ฮฑ), where the two loops are overlayed at the common AG pair. The contribution of the interaction is Ejunction+hinge(๐’ž,๐’Ÿ;ฮธ;A,G) with ฮธ the unknown pair family.
Mentions: Like ordinary secondary structure prediction tools, MC-Fold (Parisien and Major, 2008) is based on a decomposition of the RNA structure into โ€˜loopsโ€™. In contrast to the standard energy model, however, it considers the full set of base pair types available in the LW representation. Each base pair, therefore, corresponds to a triple (i,j;ฮธ) where ฮธ is one of the 12 types of pairs. In this model, ordinary secondary structures are the subset of pairs with Watson-Crick type (ฮธ=โ€˜WWโ€™) and the two nucleotides form one of the six canonical combinations {AU,UA,CG,GC,GU,UG{. This extension of the structure model also calls for a more sophisticated energy model. While the standard model assumes the contributions of the loops to be strictly additive, MC-Fold also considers interactions between adjacent unbranched loops (hairpins, stacked pairs, bulges and general interior loops). This means that the total energy of a structure is not only dependent on the loop types present, but also on the arrangement of these loops. Dispensing with details of the parametrization, the scoring function of MC-Fold for a structure ๐’ฎ on sequence x can be written as follows (see Fig. 2):(2.1)where ๐’ž,๐’žโ€ฒ,๐’žโ€ฒ are different loops of ๐’ฎ. The additive term Ec tabulates the (sequence-dependent) contributions of the loops. The interaction term Ej+h accounts for the โ€˜junctionโ€™ and โ€˜hingeโ€™ terms in stemโ€“loop regions. These interaction terms depend on the type of the adjacent loops as well as on the type ฮธ and sequence (x[k], x[l]) of the base pair that connects them. For multiloops, only the additive term is considered.Fig. 2.

Bottom Line: Successful prediction of these structural features leads to improved secondary structures with applications in tertiary structure prediction and simultaneous folding and alignment.We accompany this model with a number of programs for parameter optimization and structure prediction.All sources (optimization routines, RNA folding, RNA evaluation, extended secondary structure visualization) are published under the GPLv3 and available at www.tbi.univie.ac.at/software/rnawolf/.

View Article: PubMed Central - PubMed

Affiliation: Institute for Theoretical Chemistry, University of Vienna, A-1090 Vienna, Austria. choener@tbi.inivie.ac.at

ABSTRACT

Motivation: RNA secondary structure contains many non-canonical base pairs of different pair families. Successful prediction of these structural features leads to improved secondary structures with applications in tertiary structure prediction and simultaneous folding and alignment.

Results: We present a theoretical model capturing both RNA pair families and extended secondary structure motifs with shared nucleotides using 2-diagrams. We accompany this model with a number of programs for parameter optimization and structure prediction.

Availability: All sources (optimization routines, RNA folding, RNA evaluation, extended secondary structure visualization) are published under the GPLv3 and available at www.tbi.univie.ac.at/software/rnawolf/.

Show MeSH