Hi everyone:
I am recently reading a paper on intron retention analysis, there are several places I have not figured out myself what does it mean, one example is like this:
In order to determine if an intron retention is real from the reads number among E1I(exon1intron junction), I(intron), and IE2(intronexon2 junction), one of the criteria the author used for binomial test is:
P-value(binomial(M=min(#E1I, #IE2, #I), N=min(#E1I, #IE2, #I)+max(#E1I, #IE2, #I), P=1/3.5, alternative = lower))>=0.05
From this standard, I understand the authors want to make sure that #E1I, #I, and #IE2 have equal chance to be captured during sequence if the intron retention events is true.
What I don't understand is why set M=min(#E1I, #IE2, #I), N=min+Max, P=1/3.5 instead of M=average(#E1I, #I, #IE2), N=sum(#E1I, #I, #IE2), P=1/3
This is a statistical question, I don't know if it is OK to post it here. But if someone happened to be able to answer the question, it will be great.
Thanks for anyone who can help,
Cai
