search

chongjin / CARseq

6 stars 1 forks source link

NA p values #4

Open tannirahman39 opened 2 years ago

tannirahman39 commented 2 years ago

Hello,

I recently ran a CARseq analysis, but, I noticed that although some genes have a shrunken log fold change, the p value is NA. The log fold change is also high for these genes (e.g 2.7). My questions is: how is it possible to have a fold change but an NA p value?

Thank you!

chongjin commented 2 years ago

Hi,

Thank you for using CARseq. The result is not surprising. The p-values are calculated from a likelihood ratio statistic and the calculation involves fitting two models (a full model and a reduced model), while the log fold change is inferred from the full model only. The shrunken log fold change is estimated from a shrunken full model. Practically I would trust the p-values and shrunken log fold changes more, while the log fold change can be very unstable. Let me know if you have any questions!

On Fri, Aug 19, 2022, 11:36 AM tannirahman39 @.***> wrote:

Hello,

I recently ran a CARseq analysis, but, I noticed that although some genes have a shrunken log fold change, the p value is NA. The log fold change is also high for these genes (e.g 2.7). My questions is: how is it possible to have a fold change but an NA p value?

Thank you!

— Reply to this email directly, view it on GitHub https://github.com/Sun-lab/CARseq/issues/4, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABBIWEWST5RQPJGQ7QCRZOTVZ6SZTANCNFSM57BES2ZQ . You are receiving this because you are subscribed to this thread.Message ID: @.***>

tannirahman39 commented 2 years ago

Hello,

I may have been unclear in my previous question. I have only been using the shrunken log fold change. However, for some genes, although the shrunken log fold change is high (e.g 2.7), the p value is NA.

Thank you for your help!

chongjin commented 2 years ago

Hi,

Thanks for your clarification. As I mentioned previously, the shrunken log fold change is fitted differently, so the result does not always follow the p-values. In your example, when the shrunken log fold change and p-value results don't match, I would probably exclude the gene from the analysis.

On Fri, Aug 19, 2022, 1:31 PM tannirahman39 @.***> wrote:

Hello,

I may have been unclear in my previous question. I have only been using the shrunken log fold change. However, for some genes, although the shrunken log fold change is high (e.g 2.7), the p value is NA.

— Reply to this email directly, view it on GitHub https://github.com/Sun-lab/CARseq/issues/4#issuecomment-1220920262, or unsubscribe https://github.com/notifications/unsubscribe-auth/ABBIWETUV6LQ5WVF4JZEZK3VZ7AFZANCNFSM57BES2ZQ . You are receiving this because you commented.Message ID: @.***>