Biology Reference
In-Depth Information
calculated the hybridization energy of two RNA molecules, Rehmsmeier
et al . 131 developed RNAhybrid, a fast algorithm for predicting
miRNA/target duplexes. In essence, RNAhybrid uses the energy param-
eters from Mathews et al ., 132 which are also used by Mfold and RNAfold
secondary structure prediction programs, to compute the minimum free
energy hybrid between an miRNA and a target. Intramolecular (within-
miRNA or within-mRNA) pairs are not allowed. Important contributions
from the RNAhybrid package are the programs used to evaluate the sig-
nificance of the results. For instance, to account for the variable length of
3
UTRs, the authors use — instead of the free energy of hybridization —
normalized energies defined as with e being the free energy,
m the length of the target sequence, and n the length of the miRNA. The
statistical significance of the normalized free energies is computed using
extreme value statistics, and the significance of multiple binding sites is
computed assuming a Poisson distribution of the number of miRNA tar-
get sites per 3
e
n = log(mn)
,
e
UTR. Finally, the evolutionary conservation of putative
miRNA sites is taken into account by computing joint p -values for the
sites P [ Z 1
e 1 , Z 2
e 2 ,..., Z k
e k ]
=
(max{ P [ Z 1
e 1 ], P [ Z 2
e 2 ],...,
e k ]}) k eff , with e 1 , e 2 , ..., e k being the normalized free energies of
hybridization of the orthologous sites in the k species, and k eff being the
effective number of sequences, taking into account the phylogenetic
distance between the species. k eff is fitted using random miRNAs.
While RNAhybrid aimed to fill the need for a program that predicts
hybrids between two RNA molecules, another miRNA target prediction
program, ElMMo, 112 aimed to rigorously treat the comparative genomic
information for miRNA target prediction. With the exception of
RNAhybrid, 131 all of the previously proposed programs for miRNA tar-
get prediction that used evolutionary conservation treated each species
independently, regardless of the relative position of the species in the
phylogenetic tree. In ElMMo, the authors treated conservation in a
Bayesian framework that explicitly took the phylogeny of the species into
account, and explicitly dealt with the possibility that a conserved site is
conserved by chance in a subset of the species while being conserved due
to its being functional and under selection in another subset of species.
Thus, ElMMo infers how likely it is that a given putative site with a given
P [ Z k
Search WWH ::




Custom Search