Different FPKM values for same condition form cufflink

Question

[Prakash]

Hi all,

I want to understand why cufflink give different FPKM value for same condition (say A) when comparing with two different condition (say B and C) after analysing separately (A vs B and A vs C). Is it possible to calculate FPKM separately for each condition and then calcualate the fold change and significance in cufflink.

Thanks in advance.

[Satyajeet Khare]

I think Cufflinks calculates FPKM for each gene in transcript.gtf file that it generates. But when provided with more replicates or sample for comparison, it recalculates the false alignments based on consistency of reads in the dataset. Therefore, the FPKM changes.

You can get same FPKM for a gene in A vs B and A vs C by merging transcripts of A, B and C using Cuffmerge and then performing Cuffdiff with pairwise comparison between A & B and A & C using a single command.