Values used for calculating FPKM Value

Question

[bassanio]

Dear All,

I had got a Transcriptome Library with 10M PE reads (R1->10M & R2 ->10M). When I mapped the same to the genes from a particular genome using Bowtie2 80% of them mapped (8M).

Now I need to calculate the Expression rate for these genes. For calculating the FPKM Value I use the formula below

FPKM= MG * (10^9))/N * Length of Gene

Where;

• MG= Count of PE mapped to the genes (20,000)
• N= Library size or Mapped count ?(10M/8M?)

My question is What is the value of N should be is it 10M or 8M

Answer 1

[GouthamAtla]

It should be 8M. For example from edgeR rpkm() function, its

lib.size: library size, defaults to 'colSums(x)'

[bassanio]

If I use the same transcript data and aligned with another genome(B) where aligned reads is 9M and then if I do a DeSeq between Sample A(8M) mapped reads and Sample B(9M) mapped reads then the data is Skewed more towards Sample A as the normalization value is not same between SampleA and Sample