edgeR normalization via TMM method

Question

[seta]

Dear all,

Could you please clear me how edgeR solve the compositional bias within an RNA-seq library via TMM method? if it removes the highly expressed genes or if it allocates lower weight (via the normalization factor of less than 1) to them?

Thanks

Answer 1

[ATpoint]

StatQuest is the answer:

[seta]

Thank you for the great video. However, I have some questions, hope to find the answer.

In the video, he says “edgeR filters out the biased genes based on their log differences”, then he calculates the log2 (reference, gene 1/ sample 2, gene1). Here, I got confused about 1) why the ratio calculated relative to sample 2, not the reference sample?

2) Also, could you please tell me how “the weighted average of the log2 ratio” (step 4) was calculated?

3) Sorry, at Step 5, is the weighted average of log2 value multiplied by 2 to convert the value to normal value? What does mean “normal numbers”, here?

Thank you in advance