Number of reads to sample when using rarefy_rrna about miceco

[Russel88]

Hi Eduardo,

rarefy_rrna is very much experimental, so I don't have a clear answer for you.
But for choosing the depth, you have to be aware that rarefaction is based on the rrarefy function from vegan which is without replacement. This means that if the rarefaction depth is close to the actual read depth, the rRNA copy numbers will have very little impact on the resulting rarefied community matrix. Basically, you can never get more reads of certain taxa in the rarefied matrix. So if you want the rRNA copy number to have an effect the rarefied depth has to be significantly lower than the actual depth, and has to be that for all samples to be unbiased. Alternatively, one could add replacement when sampling reads, which should give a bigger impact of copy numbers.

It is of course problematic to choose a low rarefaction depth, as you will loose sensitivity for downstream analyses.

Best,
Jakob

[Russel88]

I have added an option to sample with replacement in the new version of MicEco 0.9.18