I´m working with small RNAs which are produced by transposons, in my data set I have different transposons with different sizes, by chance I expect that transposons with more bases could produce more small RNAs, but in my data set it's no necessary true for all transposons. I want to know if you could recommend a statistical test that help me to decide if my data are significantly different than expected in a null model.
For example, I have 1743044 small RNAs in 860753789 bases, in 4415840 bases I expect 8942 small RNAs in a null model, but in the RNA-seq experiment I found 211372, that are produced in those 4415840 bases. Mi question is, how I can know how many significant is the difference between 8942 and 211372.
You can try a exact poisson test, whereby you assume the rate of discovering a small RNA is 1743044/860753789 = 0.00202502
In R, the command would be:
poisson.test(211372,T=4415840,r=1743044/860753789)
output looks like:
Exact Poisson test
data: 211372 time base: 4415840
number of events = 211370, time base = 4415800, p-value < 2.2e-16
alternative hypothesis: true event rate is not equal to 0.00202502
95 percent confidence interval:
0.04766292 0.04807127
sample estimates:
event rate
0.04786677
The p value is extremely low.. So this means the chance of observing 211372 small RNA in 4415840 bp, given the underlying rate of 0.002 RNA per bp, is very low.
Just to be sure, I would suggest checking whether your underlying assumptions about the discovery small RNA is true, or whether your background of transposons are similar
- Badges
- Science topic
- Similar topics
- Chemistry
- Biochemistry
- Biomolecules
- Nucleic Acids
- RNA
- Small RNA