miRNA-Gene Interaction P-Values

Question

[ilikebing2000]

Hi everybody,

I searched about 'miRNA-Gene Interaction' Databases through the web to find : A database which each of it's record specifies a miRNA-Gene interaction with the respective p-value of that! I found some sort of it like TarBase, but their weren't what i look for.

Do you know any databases which I can give them a list of miRNAs and the database give me all of the genes which have interaction with the given miRNAs with the respective p-value?

Thanks.

[Lars Juhl Jensen]

If you are thinking of statistical significance for predicted miRNA targets, I think you will be disappointed. The error rates of such prediction methods are high and I would not expect that an individual prediction would ever be statistically significant (not because the methods are bad, it is just a hard problem).

Answer 1

[Ahill]

TargetScan provides a database of miRNA-gene interactions including a pCT statistic for the interactions. For human, downloadable from here: http://www.targetscan.org/cgi-bin/targetscan/data_download.vert71.cgi

[ilikebing2000]

Thanks,

Is the concept of P_CT (Probability of Conserved Targeting) similar to p-value (Probability Value in Hypothesis Testing)?

[Ahill]

Apparently no. The authors say "[the pCT] reflected the Bayesian estimate of the probability that a site is conserved due to selective maintenance of miRNA targeting rather than by chance or any other reason not pertinent to miRNA targeting". So this is definitely not a type I error measure like a traditional p-value. Higher pCT values represent higher confidence sites. Authors do describe pCT as 1 - FDR for predicting "biologically conserved target sites". I would read the methods section in this paper: https://www.ncbi.nlm.nih.gov/pubmed/18955434 for details.