Is there any established fold-change threshold for miRNA microarray data?

Question

[Joe_daisy]

I am analyzing miRNA microarray data and I would like to establish a fold-change threshold which is biologically significant. Classically for mRNA microarray a 2-fold-change cutoff was used, but I think that if we use that for miRNA as well, were could be loosing some information given that miRNA regulate gene expression. In other words, which is the miRNA expression level change, in which we could see an effect in mRNA expression? 1.5-fold change would be enough? I am asking this because in general, the fold-changes I see in the literature are not very big.

I have seen some papers not using any filtering criteria and others use 2-fold change cutoff. Is there any paper where has been studied this issue? I haven't found anything but there may be some.

Thank you

[ysora90]

I'm also very curious about it. It seems that microRNAs degrade only small amount of mRNAs (fine-tuning), but sometimes they reduce significant amount of mRNA like transcription factor. I want to find some answers for this question T^T

Answer 1

[Jean-Karim Heriche]

There's no definite answer. Depending on the process and genes one is looking at a small change can be biologically relevant whereas in another context, only a large change would have a significant biological effect. The 2-fold cutoff used is arbitrary. You could start with ranking your genes based on fold change and see if there's anything interesting towards the top and bottom. If you plot the ordered fold changes, there may be an elbow in the plot at one or both ends suggestive of cutoff values.