Limits...
A fast ab-initio method for predicting miRNA precursors in genomes.

Tempel S, Tahi F - Nucleic Acids Res. (2012)

Bottom Line: The approximation step allows a substantial decrease in the number of possibilities and thus the time required for searching.Our method was tested on different genomic sequences, and was compared with CID-miRNA, miRPara and VMir.It gives in almost all cases better sensitivity and selectivity.

View Article: PubMed Central - PubMed

Affiliation: Laboratoire IBISC, Université d'Evry-Val d'Essonne/Genopole, 23 Boulevard de France, 91034 Evry, France.

ABSTRACT
miRNAs are small non coding RNA structures which play important roles in biological processes. Finding miRNA precursors in genomes is therefore an important task, where computational methods are required. The goal of these methods is to select potential pre-miRNAs which could be validated by experimental methods. With the new generation of sequencing techniques, it is important to have fast algorithms that are able to treat whole genomes in acceptable times. We developed an algorithm based on an original method where an approximation of miRNA hairpins are first searched, before reconstituting the pre-miRNA structure. The approximation step allows a substantial decrease in the number of possibilities and thus the time required for searching. Our method was tested on different genomic sequences, and was compared with CID-miRNA, miRPara and VMir. It gives in almost all cases better sensitivity and selectivity. It is faster than CID-miRNA, miRPara and VMir: it takes ≈ 30 s to process a 1 MB sequence, when VMir takes 30 min, miRPara takes 20 h and CID-miRNA takes 55 h. We present here a fast ab-initio algorithm for searching for pre-miRNA precursors in genomes, called miRNAFold. miRNAFold is available at http://EvryRNA.ibisc.univ-evry.fr/.

Show MeSH

Related in: MedlinePlus

(A) Example of a symmetrical matrix for searching for exact and non-exact stems in a given genomic subsequence. Three stems are selected with a threshold of minimum length equal to 4 (surrounded by a blue circle). One of the three stems has been extended to a non-exact stem (surrounded by a red circle). (B) Search for hairpins. The anchor (surrounded by an orange circle) of the non-exact stem shown in (A) is positioned in the matrix, and then is extended in left and in right (green areas) on different diagonals, in order to allow bulges and internal loops.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC3367186&req=5

gks146-F2: (A) Example of a symmetrical matrix for searching for exact and non-exact stems in a given genomic subsequence. Three stems are selected with a threshold of minimum length equal to 4 (surrounded by a blue circle). One of the three stems has been extended to a non-exact stem (surrounded by a red circle). (B) Search for hairpins. The anchor (surrounded by an orange circle) of the non-exact stem shown in (A) is positioned in the matrix, and then is extended in left and in right (green areas) on different diagonals, in order to allow bulges and internal loops.

Mentions: Stems of length greater than a minimal size lmin are searched for in the matrix. For example in Figure 2A, three stems (surrounded by blue) are selected. The 10 longest stems verifying a certain percentage of criteria [set by default to 70% (see ‘Results’ section or given by the user)] are then selected. The 12 exact stem criteria are section the size of the exact-stem, the MFE, the size of the terminal loop, the percentage of A, C, G and U, the number of consecutive A, C, G and U and the ratio of GU pairings in the stem (Supplementary File 1).Figure 2.


A fast ab-initio method for predicting miRNA precursors in genomes.

Tempel S, Tahi F - Nucleic Acids Res. (2012)

(A) Example of a symmetrical matrix for searching for exact and non-exact stems in a given genomic subsequence. Three stems are selected with a threshold of minimum length equal to 4 (surrounded by a blue circle). One of the three stems has been extended to a non-exact stem (surrounded by a red circle). (B) Search for hairpins. The anchor (surrounded by an orange circle) of the non-exact stem shown in (A) is positioned in the matrix, and then is extended in left and in right (green areas) on different diagonals, in order to allow bulges and internal loops.
© Copyright Policy - creative-commons
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC3367186&req=5

gks146-F2: (A) Example of a symmetrical matrix for searching for exact and non-exact stems in a given genomic subsequence. Three stems are selected with a threshold of minimum length equal to 4 (surrounded by a blue circle). One of the three stems has been extended to a non-exact stem (surrounded by a red circle). (B) Search for hairpins. The anchor (surrounded by an orange circle) of the non-exact stem shown in (A) is positioned in the matrix, and then is extended in left and in right (green areas) on different diagonals, in order to allow bulges and internal loops.
Mentions: Stems of length greater than a minimal size lmin are searched for in the matrix. For example in Figure 2A, three stems (surrounded by blue) are selected. The 10 longest stems verifying a certain percentage of criteria [set by default to 70% (see ‘Results’ section or given by the user)] are then selected. The 12 exact stem criteria are section the size of the exact-stem, the MFE, the size of the terminal loop, the percentage of A, C, G and U, the number of consecutive A, C, G and U and the ratio of GU pairings in the stem (Supplementary File 1).Figure 2.

Bottom Line: The approximation step allows a substantial decrease in the number of possibilities and thus the time required for searching.Our method was tested on different genomic sequences, and was compared with CID-miRNA, miRPara and VMir.It gives in almost all cases better sensitivity and selectivity.

View Article: PubMed Central - PubMed

Affiliation: Laboratoire IBISC, Université d'Evry-Val d'Essonne/Genopole, 23 Boulevard de France, 91034 Evry, France.

ABSTRACT
miRNAs are small non coding RNA structures which play important roles in biological processes. Finding miRNA precursors in genomes is therefore an important task, where computational methods are required. The goal of these methods is to select potential pre-miRNAs which could be validated by experimental methods. With the new generation of sequencing techniques, it is important to have fast algorithms that are able to treat whole genomes in acceptable times. We developed an algorithm based on an original method where an approximation of miRNA hairpins are first searched, before reconstituting the pre-miRNA structure. The approximation step allows a substantial decrease in the number of possibilities and thus the time required for searching. Our method was tested on different genomic sequences, and was compared with CID-miRNA, miRPara and VMir. It gives in almost all cases better sensitivity and selectivity. It is faster than CID-miRNA, miRPara and VMir: it takes ≈ 30 s to process a 1 MB sequence, when VMir takes 30 min, miRPara takes 20 h and CID-miRNA takes 55 h. We present here a fast ab-initio algorithm for searching for pre-miRNA precursors in genomes, called miRNAFold. miRNAFold is available at http://EvryRNA.ibisc.univ-evry.fr/.

Show MeSH
Related in: MedlinePlus