Is it possible to use a self made txt doc containing normalised FPKM values in cuffdiff - for the purpose of multiple testing (calculate FDR / q value)

Question

[sangita_b]

I have used the Tuxedo pipeline to analyse my reads (RNASeq) for differential expression. The design of my experiment is: 5 conditiona (3 gene variants in an overexpression plasmid, empty vector control, and untransfected) replicated twice in two donors. The cells are primary human cells. After running cuffdiff I do not see many genes that meet FDR of 5% and very few that that are 10%. This may be due to donor (inter and intra) variation. One suggestion made was to normalise the data to the untransfected/ media control which I have done in excel. I'd like to know if there is a programme that I can use to carry out multiple testing on this data/ or if I can use this file in cuffdiff (for FDR/ to generate q values) so that I can interpret my data with confidence. An alternative pipeline (htseqcount_DESEQ) has been suggested which I am also following up. I've been advised that the model used in DESEQ may be better suited to my experimental design compared to Cuffdiff.

[Devon Ryan]

In order to do this with cuffdiff you'd probably have to make your own CXB files, which would require writing a program to do that. You're life will be easier with DESeq2 or edgeR or limma/voom. Having said that, whether even these will work depends a bit on exactly what sort of normalization you're trying to do. Providing some details there might allow better feedback.

[maria]

sorry how did you exclude the over expression effect from the plasmid?