Normalization necessary when converting to relative frequencies?

[njnealon (Nora Jean Nealon)]

Hello! Just a follow-up to this... I downloaded the .csv files from my taxa barplots visualization (for each of the seven levels of taxonomies). I agree with hand-calculating relative frequencies. However, I just want to make sure I am doing that calculation correctly.

Each of my samples has a different total number of reads associated with it. So for each sample, and each OTU in that sample I will do the following: (# reads in the OTU/# reads in the entire sample) = relative frequency of the given OTU in the given sample.

I'm really new to this type of analysis and my concern was that I should be using some type of normalized reads across all samples before calculating frequencies of individual OTU's across samples?

Thank you!

[Nicholas_Bokulich]

Hi @njnealon,
The relative-frequency method will perform this calculation for you.

You can also collapse to the taxonomic level of choice before calculating relative abundance, if that's your goal.

But yes, the calculation that you describe would be the way to do this by hand.

There is no need to do any additional normalization steps prior to calculating relative frequencies.

I hope that helps!