Help resolve a lab debate...

[lre1234]

Hi all,
We are having a lab debate regarding short RNA sequence. Specifically, we generally we do unique mapping of the reads to the genome. After intersecting the mapped reads to the miRNAs and other short RNAs we want to calculate a RPM value. My lab mate wants to normalize the expression to the total number of raw reads (total off of the machine). Me, on the other hand thinks that we should be normailzing to the total number of mapped reads.

Can someone possible way in on this...

[dpryan]

You're both wrong, you should normalize all of the samples together using a more robust statistic (e.g., TMM), rather than something problematic like RPM.

However, if you must use RPM for some reason, then the correct method would be to use the total number of aligned reads. You don't want differences in alignment rate to change your metrics (though you should pay attention to these for quality control purposes).