Hi, we are analyzing our first RNAseq experiment and in order to understand if in the future we could afford to multiplex our samples we wanted to check in with one fourth of the reads we would find results superimposable with all of them.
As long as the alignment step is slow we though to start by subsampling the accepted_hits.bam...right now I've selected randomly one line over four of the derived sam and analyzed the results with cuffdiff.
Sperman correlation between FPKM values are good and the overlap between up and downregulated genes (cutting at FDR 0.05) is significant (accordint to the hypergeometric test) but I'm puzzled because the number of downregulated genes is higher in the "subsampled" case: 3459 vs 201.
Am I missing something?
As long as the alignment step is slow we though to start by subsampling the accepted_hits.bam...right now I've selected randomly one line over four of the derived sam and analyzed the results with cuffdiff.
Sperman correlation between FPKM values are good and the overlap between up and downregulated genes (cutting at FDR 0.05) is significant (accordint to the hypergeometric test) but I'm puzzled because the number of downregulated genes is higher in the "subsampled" case: 3459 vs 201.
Am I missing something?