For 242 site functional effect is not median

[Bernadetadad]

Something weird is going on with this site 242 - the median effect is for some reason 1.12 when it should be 1.0295 based on library replicate values. But it seems like this is only an issue with this site all other sites I looked at are calculated properly, not sure how is this possible

[jbloom]

@Bernadetadad, the way the analysis currently works is that the median is across all replicates, not across the two libraries.

Here are the per-replicate values:

The median of those is actually 1.12.

You could also argue instead the approach should first take the median of each library, and then take the median of that. Doing so would give the result of 1.0295 that you are expecting.

However, I think current approach may be better because within and between replicates values are similarly correlated, suggesting to me that it's better to treat each as it's own data point: