Biomedical Engineering Reference
In-Depth Information
for instance a complimentary nuclei the free energy of binding must be within 33 %
of the optimum free energy, while imperfect nuclei must be within 66 %. The re-
maining results are scored against a Hidden Markov Model (HMM) based ranking
of the probability of being common targets of subsets of the miRNAs under con-
sideration. Finally conservation analysis across eight vertebrate species is utilised
to prioritise targets and eliminate false positives. The scores for each species are
calculated individually and combined to give a final score for a gene. The usefulness
of this algorithm is demonstrated by PicTar prediction and subsequent experimental
confirmation of miR-375, miR-124 and let-7b coregulation of the myotrophin gene
(Krek et al. 2005 ).
3.3.3
Diana-microT
The Diana-microT (Maragkakis et al. 2009a , b ) approach utilises an extended miRNA
seed region of up to nine nucleotides known as the “miRNA driver sequence”. This
driver sequence is shifted across the 3 UTR of the target gene to search for alignments
with 7, 8 or 9 nt Watson-Crick base-pairing. DIANA-microT also aims to incorpo-
rate sites with weaker binding of 6 nts or G:U wobbles within target prediction.
DIANA-microT 3.0 attempts to control false positives by filtering potential target
sites without 7 nt Watson-Crick base pairing by applying a threshold to thermody-
namic calculations provided by RNAhybrid (See below). Each individual miRNA
binding site is scored with respect to the binding type (e.g. 6mer) and conservation.
These scores are combined for each individual 3 UTR to produce a gene level “miTG
score” (incorporating assessment against a set of synthetic miRNAs).
3.3.4
TargetScan/TargetScanS
TargetScan was the first method developed to predict miRNA targets in humans
(Lewis et al. 2003 ). The underlying rationale of the TargetScan method is that, by
returning only perfect seed matches and utilising only groups of orthologous 3 UTRs
as input, the false positive rate can be controlled (although non-conserved sites will
be missed). The algorithm begins by locating perfect complementarity to the miRNA
seed regions. Once a seed match is located TargetScan extends the alignment along
the 3 UTR until a mismatch is located (G:U wobbles are permitted). The optimum
base pairing at the 3 end of the miRNA is determined and RNAfold calculates the
free energy to the miRNA-mRNA interaction. For each alignment a “Z-score” incor-
porating thermodynamic stability is calculated for each species to assign a rank to
each potential miRNA-target interaction. Further refinement produced the simplified
TargetScanS (Lewis et al. 2005 ) which is a simplification of TargetScan algorithm
where the criteria for a target match is less stringent and only complementarity to
a 6 nt seed match along with an adenosine located at position 1 is required. The
Search WWH ::




Custom Search