Maximizing statistical power with limited samples I'm designing an experiment where I observe gene expression for cells that either have a purturbation, or nothing at all (control). 
The hope is to analyze as many perturbations as possible.
Since I intend on comparing the perturbation samples to the control samples, my intuition is that I get a lot of value from doing many repeats of the control samples (since I will be using them in every test)
Can I get away with doing a bunch of repeats for the control and a single observation for each condition. For example, if I have enough resources to do 100 experiments, do the following:


*

*10 X control experiments

*1 X perturbation 1

*1 X perturbation 2

*...

*1 X perturbation 90 

 A: This article may be useful to you: 
Bruhn, Miriam, and David McKenzie. "In pursuit of balance: Randomization in practice in development field experiments." American economic journal: applied economics 1.4 (2009): 200-232.
A: You never want to have just one case per treatment, as you then can't even judge whether some unexpected technical glitch is responsible for a particular value. Most experimental designs work best if you have the same numbers of cases in each group, including the control group. ANOVA then efficiently pools information about the experimental errors from all cases/treatments, providing more reliable estimates of the actual treatment-related effects and their likely errors in estimation.
For cell-line work in particular I would, as a reviewer, want to see at least 3 separate biological replicates for each set of treatments/control. For example, grow up a large dish of cells, separate into smaller dishes for treatments/control testing. Then repeat the process two more times starting with a fresh vial of the cell line each time.
A: It may be a better idea to utilize a loop design to balance your replicates across treatment categories and then fit a MAANOVA model. 
An introduction to this topic is the MAANOVA paper/software package by Kerr and Churchill. PDF here: http://lectures.molgen.mpg.de/Genexpression_WS0506/material/wu2002.pdf
