Hello,

Here: https://imgur.com/z7Fkyjv you can find the PCA of the experiment that I am analyzing. Cond1 e cond2 were designed to have an over expression of two different genes. The sample cond1 near to controls has an expression of that gene much less than the others biological replicates. In this case I know that in that sample something did not work correctly and is like to have another control. Is it advisable to remove that sample and to have only two replicates for cond1 or is it better to have another replicate but a bad replicate in which the experiment did not work?

Thank you. 

If you know the sample failed (technical problem) then do not include it.