How to normalize intron counts?

Question

[PK]

Hi all,

I extracted the intron counts for the genes. After the Deseq, i looked at the differentially expressed gene manually. I observed strangely second and third introns has more read compare to rest of the introns. It is pretty clear that if i take these genes intron and compared just raw counts i could see the difference. But i just don't know how to normalize this with respect to wildtype. I did very primitive way

KO_repA-Intron-1/ KO_total_counts_of_repA

But this approach completely shows negative effect on intorn-2 and 3. I know why this is happening because i have many counts in the KO samples so sum value also higher compare to wild type. Will RPM help? I don't want to sum against the gene.

[Asaf]

I think you can apply a chi-square test here with the table being 2X3: KO and WT are rows, introns 1,2,3 are columns. If the reads are divided differently between the introns in WT vs KO you'll see it in the test. If introns 2 and 3 have more reads in both KO and WT then you don't need to normalize, just show that the numbers are higher in all samples.