GEO studies included very few subjects having extreme statistical power

Question

[786]

Hi everyone! I couldn't understand this I need your help that why some GEO studies included very few subjects and however have an extreme statistical power (a lot of genes significant)?? Please help me in understanding this.

Answer 1

[Asaf]

The reason of having replicates is to be able to distinguish variability between conditions and within conditions and would be able to tell when the difference you see between two measurements stems from expected variation that you might observe between two biological replicates or it's truly an effect of the different conditions. In order to get real statistical power, that would apply for a t-test, you would need more than 10 repeats, which is almost never done. Statistical analysis tools of microarray/RNAseq use tricks to enhance the statistical power of a few replicates. One such a trick is to assume that the within condition variability (biological "noise") is similar among genes with the same expression level. Using this assumption the standard deviation of the expression level in each condition can be better evaluated and the test is stronger. That's the concept, not the exact implementation.

Answer 2

[12021560-002]

it seems difficult to understand, it is humbly requested to explain it a comprehensive way.