I am trying to integrate expression data of miRNA-mRNA (~20 samples). I have a few queries such as which one is the best method for co-expression calculation (my interest is anticorrelation) 'Pearson', 'spearman','Kendall', 'Mutual information' etc? and why?
My aim is to extract miRNA-mRNA pairs with anticorrelation <-0.3 with statistical significance and further mapped those pairs with different databases which contains interactions either predicted or experimentally validated. One R Package Anamir seems to have these functionalities but it misses the significance part (p-value of the correlations).
I did above part by my own, however, I am sure what is the best input for calculation of correlation; eg, the log transformation of raw values or cpm value or other normalized expression values?
It would be great if anyone knows any good package or this kind of analysis or can explain some of my queries.