Hello there, 

I've been having some heated discussions about whether this scenario using a cell line (passaged over and over) constitutes a paired analysis or not. By this point, I thought it was wise to ask the community to get the insights of experts.

Consider growing a cell line three different flasks assuming to be a biological replicate each. So three bioreps. You then split each flask across multiple plates to grow the cells for drug exposures (say 3 6-well plates). 

Drug exposures are as follows: 1mg, 3mg and 10mg of drug A, same for drug B and drug C. You keep untreated control wells too. 

This is what you end up with:

Drug A 1mg - 3 samples / Drug A 3mg - 3 samples / Drug A 10mg - 3 samples (same for B and C and 3 controls). 

You then analyse the drugs separately, each concentration to the controls hence:

Drug A biorep 1, biorep 2, biorep 3 versus Control biorep 1, biorep 2 and biorep 3. 

Are these paired? Technically I feel this should be a block design whereby Drug A biorep 1 would be related to Control biorep 1 because these come from the same flask. However, the convention in statistics seems to be to just analyse all treatment independently unless there's a time factor involved (btw time here is the same i.e. say 48h for all). 

Others have argued that each well constitutes its own new entity, but given that were are using the same cell line that seems a bit far fetched. 

Anyone got any ideas?

Thank you very much.