module-traitRelationships negative correlation p values incorrect

[richard-k-perez]

Hello,

There is an issue with p value calculation for negative correlations in the module-trait relationship. For example, if I calculate the correlation A vs. B, the results are R=0.7 p <0.01. When I use the same data but now make the trait B vs. A, the results are R =-0.7 p = 0.99. The p-value is always 1-p of the opposite comparison. I tested this myself by extracting the eigenvalues for each of my samples for a module and switched the order of comparison. I reproduced the correlation for both comparisons but only the p value for the correlation. The p value for the negative correlation in my test, was as significant as the positive correlation, which is to be expected.

Hoping you could look into this! May be a very quick fix.

[nargesr]

Hi,

Thanks for using PyWGCNA, would you mind telling me the approach you used?

On my side, I used scipy.stats.pearsonr with alternative='greater', so I would assume your observation it because you probably go with the alternative='two-sided'.

I'm happy to change the alternative parameter if that's the problem here

[richard-k-perez]

Hello, Ah I think that’s the issue! Yeah I used the default for the Pearson r test, which happens to be the two sided. I can check later tonight or tomorrow if I can recreate the issue but I think it’s like you said. Could you change the alternative parameter to two-sided or make it an option? Thanks, Richard

…

On Fri, Oct 27, 2023 at 13:50 Narges Rezaie ***@***.***> wrote: Hi, Thanks for using PyWGCNA, would you mind telling me the approach you used? On my side, I used scipy.stats.pearsonr <https://docs.scipy.org/doc/scipy/reference/generated/scipy.stats.pearsonr.html> with alternative='greater', so I would assume your observation it because you probably go with the alternative='two-sided'. I'm happy to change the alternative parameter if that's the problem here — Reply to this email directly, view it on GitHub <#73 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AHBOUAGEUWLTDW6LJZKVN23YBQM73AVCNFSM6AAAAAA6TJO5GGVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTOOBTGQ4TOMBRGY> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[nargesr]

Hi @richard-k-perez

sure happy to change it and make it an option in the next release.

Will close this issue once I make a new release.