JSB-UCLA
commented
2 years ago Hi JoWhi,
Thank you for your question. We will add a functionality to GitHub to check if the Clipper assumptions hold. Please stay tuned for our package update.
For now, a rough check you may do is to look at the histogram of the contrast scores Clipper outputs. If the Clipper assumptions hold, the histogram should be largely symmetric around 0 with the right tail heavier than the left tail. The reason is that Clipper’s contrast score construction is to ensure that null features’ contrast scores are symmetric around 0, and non-null features’ contrast scores are largely positive.
Please let us know if you have more questions.
Best, Jessica
On Tue, Apr 5, 2022 at 10:41 AM JoWhi @.***> wrote:
Hi, I've run Clipper for a differential methylation analysis on a small cohort, and now I'm wondering how to verify that this was an appropriate fit for my data. Normally I'd look at a p-value histogram and QQ plot to assess a linear model fit, but without p values, I'm not sure how to proceed. Thanks for any suggestions!
— Reply to this email directly, view it on GitHub https://github.com/JSB-UCLA/Clipper/issues/1, or unsubscribe https://github.com/notifications/unsubscribe-auth/AQZP7H4FACOJ6WTEMSWRBSTVDR3NBANCNFSM5STQCJUQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>
-- Best, Jessica
Jingyi Jessica Li, Ph.D.
Associate Professor Department of Statistics University of California, Los Angeles
Hi, I've run Clipper for a differential methylation analysis on a small cohort, and now I'm wondering how to verify that this was an appropriate fit for my data. Normally I'd look at a p-value histogram and QQ plot to assess a linear model fit, but without p values, I'm not sure how to proceed. Thanks for any suggestions!