(B ) FOXL2 peak enrichment at the proximity of in vivo FOXL2 targets (i.e., intersection of targets identified in our data with genes whose expression is modified following the Foxl2 conditional knockout, Uhlenhaut et al. 2009 ).
Fig. 5 Intersection of drug-target and target-adverse effect data (a).
As an alternative to prediction by miRanda (Enright et al., 2003) alone, we used the intersection set of target predictions by PITA (Kertesz et al., 2007), DIANA-microT-v4 (Reczko et al., 2012) and seed matching using canonical seeds (Bartel, 2009).
EDP and OVS performed the intersection of potential target genes and known pharmacological targets.
Furthermore, the intersection of the target gene sets derived from the miRBase versus Made1 consists of a mere 29 genes, and this figure is only slightly higher than the random expectation of 27 shared targets (P = 0.07 binomial distribution).
The probability to obtain an intersection size | I t| is computed by the formula, where x represents the random variable for the intersection of two target sets.
For each combination of disease and TF (i.e. the Cartesian product), we find the intersection of TF target genes and disease-associated genes across the genome.
p (x = | I t | ) = (| E t | | I t | ) (| G | − | E t | | B t | − | I t | ) (| G | | B t | ) The P-value P t as the probability of observing an intersection this large or greater can thus be computed by the formula, where x represents the random variable for the intersection of two target sets.
Final lists of putative targets are generated by the intersection of the targets predicted from TargetScan and miRBase.
For miRNAs that activated signaling in our screen, we took the intersection of predicted targets and negative regulators from the cDNA screen.
