If during the library preparation I make different PCR cycles will my differential expression analysis will be biased because of the samples with higher PCR cycles? Would it be noiser ?
Yes, you might detect PCR bias as DE genes.
So then all the samples should have the same PCR cycles? But why they will be biased if during the differential expression analysis I normalized the samples?
I depends a bit on your experimental design meaning do you have 5 cases and 5 controls and all controls have more PCR cycles then yes it might be an issue. So maybe elaborate a bit more about the project and planned analyses...