How to look at the global level of miRNA after performing DEA for miRNA-Seq data?

Question

[Azade]

Hi all,

I want to look at the global level of cellular miRNA after a treatment. After performing DEA for miRNA-Seq data, is it correct to count the number of up-regulated and down-regulated miRNAs, then if the number of up-regulated more, I conclude that the global level of cellular miRNA becomes up?

Thanks for any help.

Answer 1

[Asaf]

You can't do that. If, for instance two miRNAs are at the same level then one miRNA is up 10 fold and the other is down 2 fold you'll end with more miRNA. One way you can by proxy estimate the number of miRNA is by looking at the miRNA machinery (the nucleases, chaperons etc.) if they go down in one condition relative to another condition then you can assume you have less miRNAs in condition A relative to B since miRNA usually have a short half-life.

[Azade]

Thanks for your informative comment.
How about the normalized counts? for example, I have two groups(tumor and normal) and I want to compare the miRNA pool of them. First I create a pseudo sample column for each group that has the mean of normalized counts of each miRNA within that group, then calculating the sum of values in pseudo sample columns, and finally, comparing them.

Something like this:

    Normal  Normal  Normal  Normal  Normal  Normal  Tumor   Tumor   Tumor   Tumor   Tumor   Tumor   Tumor   Tumor   Tumor   Tumor   Pseudo_Normal   Pseudo_Tumor
hsa-mir-1287    6.429984182 7.917527332 5.460891284 6.13848406  5.63643575  6.885285575 6.429085183 4.835015014 5.614392282 3.849006762 5.511832367 6.477928554 6.539638176 5.984423651 5.963477183 6.687337593 6.4114  5.7892
hsa-mir-1288    2.23428809  1.147984241 1.98862615  0   0   1.026667424 1.253898695 1.339201049 1.909965408 0   1.185912222 1.303883211 1.312879736 2.309282049 2.401556791 1.17632103  1.0662  1.4192
hsa-mir-1289-1  0   0   0   0   0   0   0   0   0   0   0   0   1.312879736 0   0   0   0   0.1312
hsa-mir-129-1   2.23428809  1.779083593 5.950843362 5.840691805 3.948835807 5.833266716 2.709158517 1.339201049 2.436375461 4.007764538 4.419616583 1.977408407 1.988680656 0.992109221 2.401556791 1.815573192 4.2645  2.4087
hsa-mir-129-2   1.512303888 1.147984241 6.018470894 6.13848406  5.026490293 6.300095377 2.854360015 0   2.576377401 3.353243138 4.503108551 2.434768593 3.509770089 0   0   2.594545942 4.3573  2.1826
hsa-mir-1291    1.917996264 1.779083593 0   0   1.48816244  1.026667424 0.75909202  1.339201049 0.633756846 0   1.185912222 0.794488791 2.447078826 1.574431842 2.401556791 0   1.0353  1.1113
hsa-mir-1292    1.917996264 1.147984241 0   1.524549825 0   1.026667424 1.914449927 2.831967729 0   0.878238961 1.185912222 1.303883211 2.794411255 2.571787123 1.279714976 3.097608037 0.9361  1.7857

And compare the sum of Pseudo_Normal column values with Pseudo_Tumor column values.

[Asaf]

The amount of DNA that goes into the sequencer is equal between samples, it doesn't matter how much RNA was there to begin with, it will even out so all samples will be equally represented. So more reads doesn't mean more RNA. That's in addition to all the bias introduced when building the library, and there's a lot of it especially when talking about miRNA.