testing similarity of covarege for pair of genes / Fisher test? / contingency table

Question

[agata]

I would like to know whether the pair of chosen genes have similar coverage. To do so, I think it will be good to apply the Fisher test (maybe not?) However, I have a problem with creating a contingency table.

From summary featureCounts table (before normalisation) I took the number of all aligned reads in my mutant: 14340632+15907088+18183032=48430752. The number of aligned reads in gene X region: 6+15+1=22 The number of aligned reads in gene Y region: 27+30+32=86 (I have data with 3 biological replicas, so I simply sum up numbers; I took data before normalisation with DeSeq2).

I guess that the table should look somehow like that:

But I sincerely do not know how to fill it properly and I am probably missing something important in my assumptions, because when I will put 48430752 - 22, and then to the next cell 48430752 - 86 it will not sum up correctly.

Answer 1

[agata]

I gave up idea of applying Fisher test after all. Instead I applied the machine learning solution: I plotted all genes with the number of reads and clustered it by k-means, without defining the number of clusters. After all I got two clusters and the pair of gene that I was interested in was placed in the same place on a plot.