we have a data set of two conditions with each three replicate. in our data ,we assume that there is a transcriptional amplification toward one of the conditions and therefore we used the ercc spikeIn in the data set.
To normalize the data set on the spikeIn level I have followed the method from the paper here.
"We used a loess regression to renormalize the RPKM values by using only the spike-in values to fit the loess. The affy package in R provides a function, loess.normalize, which will perform loess regression on a matrix of values ... Eighteen thousand five hundred and thirty-six genes with a RPKM value of 1.0 or greater in the low-Myc sample were selected, and the log2 fold ratio between the low-Myc and high-Myc samples were calculated and shown as a heatmap."
But is this the way to find significant genes? Just taking the renormalize values>=1?
To re-calculate the significance - Can I just calculate a t-test p-value and a FDR q-value?
What other methods do I have to calculate the new significance?