search

4ureliek / TEanalysis

Analysis of TE contribution to features (transcripts or simple features). Includes utils to test enrichment.
MIT License
23 stars 11 forks source link

How to evaluate significance based on binomial_test？ #7

Closed woshiyangsi closed 2 years ago

woshiyangsi commented 2 years ago

Hi @4ureliek,

I tried to evaluate the significance based on two tailed permutation test, so I used rank is > 5 and pvalue is not "ns" to filter, but I did not get significant TEs. Therefore, I want to filter TE based on binomial_test, which conditions should I filter based on at this time? Should I filter directly based on p value?

Looking forward to your reply, Siyu

4ureliek commented 2 years ago

Hi, I am not a strong statistician - I would tend to think that if your permutation pvalues are not significant, then you might want to try more bootstraps? Otherwise yes you can filter on binomial_test pvalues, but they may be less reliable.