This is a biological problem, however the problem requires a statistical solution, so therefore I'll keep it as abstract as possible in order for statisticians to be also able to contemplate it - I'm also only partially sure it requires a conditional probability solution.

Basically, I have a bunch of 'miRNAs' and these target certain proteins:

miRNA1 ----> protein1
miRNA1 ----> protein2
miRNA1 ----> protein3
miRNA1 ----> protein4
miRNA1 ----> protein5

miRNA2 ----> protein3
miRNA2 ----> protein4
miRNA2 ----> protein5

There are many different miRNAs that target many different proteins.

However, up to two different miRNAs come from the same 'area'.

Therefore I want to determine whether given that these miRNAs come from the same area: are their gene targets more similar?

Any help on this matter would be extremely useful!

  • 2
    $\begingroup$ Do you have a specific method in mind for measuring similarity between two sets of gene targets, or is that part of the question? $\endgroup$
    – DavidR
    Mar 23, 2014 at 13:55
  • $\begingroup$ That really is part of the question: I don't know how you'd go about measuring their similarity. What methods are there for measuring how similar these targets are? $\endgroup$ Mar 23, 2014 at 17:10

1 Answer 1


Here's one way you might want to do it: take all pairs of miRNAs, and for each pair calculate what fraction of the proteins that are targeted by either miRNA are targeted by both miRNAs. Now each pair is associated with a number between 0 and 1, and you can see how the subset of these numbers that come from pairs from the same "area" compare to the overall set.

  • $\begingroup$ Hi Daniel, I see you've just joined, welcome! So this is to see if the pairs have a higher propensity to bind to the same protein as compared to that protein across all pairs? I'm a bit confused as to what you mean by the overall set. $\endgroup$ Apr 20, 2014 at 14:15
  • $\begingroup$ I was saying that you could, as a first step, compare two histograms: the overlap fractions of all pairs of miRNAs, and the overlap fractions of pairs of miRNAs from the same area. If these are different, that would just tell you that the areas made some overall difference. Does that make sense? $\endgroup$ Apr 20, 2014 at 16:26
  • $\begingroup$ Ahh i see, so it's kind of like utilising fuzzy set theory. This is the approach I eventually went with, or a very similar approach anyway. I call this question closed, it's pretty useless to most people out there! $\endgroup$ Apr 20, 2014 at 23:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.