I have got time-course RNAseq experiments for three time points: 0h, 1h and 6h. It's done for five strains (3x5 samples) in the same contidions and for one strain for five conditions (3x5x1 samples). But replicates were done only for one strain and for one condition.
I have noticed, cufflinks detects >10x more genes as significantly differentially expressed when replicate are used as compared to same data but with only one replicate (12136 both replicates, 1115 rep1, and 923 rep2). This complicates my analysis a lot...
Can as use those replicates to estimate confidence values of DE genes for other strains/conditions with no replicate?
Or maybe there is a way to re-sample data with no replicate in order to simulate replicates?
Whatever will work better...