confused about Psi_95CI
QuanlongJiang opened this issue · 1 comments
QuanlongJiang commented
Dear author,
- I think the psi 95CI shuld be a invterval. but only a single value was provied in the result. Is Psi_95CI the upper bounds of 95 CI?
- you said " if a splicing event in a gene doesn’t have any read, it will return a posterior with Psi’s mean=0.5 and 95% confidence interval around 0.95 (most cases >0.9)" . I think PSI 0.5 means a half of exon was exclued. This is confused. Is that better if the PSI assigned as 1 if there is no reads?
- why Psi_95CI > 0.3 keep the confident psi?
Thanks a lot, I hope to hear from you.
quanlong Jiang
huangyh09 commented
Hi, thanks for the questions.
- You are right, 95 CI refers to the lower and upper bound of 95% interval. Here, we compact it to a single value to denote the range (upper - lower bounds), as we aim to use for filtering.
- that's the impact of how we choose the prior. In BRIE2, it is recommended to use the average of the cell population as the mean of the prior (a logit-normal distribution). Nonetheless, one should keep an eye on the 95 CI to filter out those less confident quantifications.
- It should be Psi_95CI < 0.3 as confident psi. One can choose a more stringent cutoff, if they are more conservative or have very high coverage.
Yuanhua