mirna / rna-seq integration

Question

[ystokar]

I have RNA-SEQ data as well as miRNA-seq from the same samples. I'm looking for the best way to integrate the two. One option i thought of is to use a target prediction tool on the differently expressed MIRs and then check for negative correlations in expressed RNA data. Any suggestions?

[shawn.w.foley]

What's the study design and ultimate goal? Are you comparing miRNA and RNA levels in treated vs untreated cells? Different cell types? Developmental phases? Are you trying to define miRNA targets?

I'd be mindful about assuming you should see negative correlations depending on what organism these data are from. In many plants you'll see miRNAs leading to degradation of the target, so you're assumption is sound, but in many animals (humans especially) you'll often see miRNAs functioning to inhibit translation, so you'd expect an anti-correlation between the protein level and the miRNA.

It's also worth noting that each miRNA has dozens to hundreds of targets, so it doesn't take very many miRNAs to make your "potential" targets almost genome-wide. If you have additional publicly available data such as CLIP-seq of AGO or ribosome footprinting then you could use these data to draw more meaningful conclusions about miRNA-target relationships.

[ystokar]

Thanks. see my answer below

[ystokar]

study design is analysis of liver RNA from treated vs untreated groups of mice. Goal is to find novel mediators of change on transcriptional level.

[shoujun.gu]

There is no way to know which integration method is the best, unless you have a ground truth validation method after integration.