5.0 years ago by
Australia, ACT, Canberra, ANU
Based on the fact that "cuffnorm" generates normalized expression values on the various tracked features (gene/transcript/TSS/CDS), it does not make sense to use the cuffnorm output as input for further analysis in DESeq. The "cuffnorm" manual entry specifically states:
"Cuffnorm will report both FPKM values and normalized, estimates for the number of fragments that originate from each gene, transcript, TSS group, and CDS group. Note that because these counts are already normalized to account for differences in library size, they should not be used with downstream differential expression tools that require raw counts as input."
The reason for this is, that cufflinks/cuffnorm use specific models for estimating transcript abundance, which are producing results that are not compatible with count-based types of analysis such as used in DESeq and edgeR. Both DESeq and edgeR require tables of raw counts in order to estimate the parameters for their models (dispersion).
The only way to run DESeq would be by creating count tables for each sample using e.g. htseq-count or methods from the GenomicRanges/IRanges packages, such as findOverlaps.
A good methods overview paper was recently published here: http://www.nature.com/nmeth/journal/v12/n2/full/nmeth.3252.html [a bit silly though putting a paper on Open Source software behind a paywall...]
Hope this helps!