Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Question on adjusted P values reported in collapsed output #51

Open
mt1022 opened this issue Aug 16, 2022 · 2 comments
Open

Question on adjusted P values reported in collapsed output #51

mt1022 opened this issue Aug 16, 2022 · 2 comments

Comments

@mt1022
Copy link

mt1022 commented Aug 16, 2022

Dear developers,

It seems that the adjusted p values in the collapsed output was calculated from filtered p values. Shouldn't multiple testing correction be done before filtering p-values with the default pvalue cutoff of 0.05? To do the multiple testing correction correctly, could I set --pval-cutoff 1 and then filter the output by adjusted_pval < 0.05 manually?

> collapsed[order(-pval_combined)]
      pval_combined adjusted_pval
    1:  4.993873e-02  4.993873e-02
    2:  4.977925e-02  4.978278e-02
    3:  4.968042e-02  4.968748e-02
    4:  4.958320e-02  4.959377e-02
    5:  4.957438e-02  4.958848e-02
   ---                            
14073: 6.443817e-267 1.814192e-263
14074: 3.203795e-276 1.127496e-272
14075:  0.000000e+00  0.000000e+00
14076:  0.000000e+00  0.000000e+00
14077:  0.000000e+00  0.000000e+00

Besides, I noticed that there are two parameters on calculating combined p-values:

  --dependence_test {none,mic,pcc}
                        the method for measuring the dependence between frame1 and frame2. This test could help determine whether the combined p-values should be ajusted to account for the dependence between two test (i.e. F0 vs F1 and F0 vs F2). mic: Maximal Information
                        Coefficient; pcc: Pearson Correlation Coefficient.
  --stouffer_adj {none,nyholt,liji,gao,galwey}
                        the method for adjustment the cominbed p-values to account for the dependence between two tests (i.e. F0 vs F1 and F0 vs F2). see details at: https://search.r-project.org/CRAN/refmans/poolr/html/stouffer.html

What's the recommended way to set the two parameters?

Thanks in advance.

@ericmalekos
Copy link

Hi,
@mt1022 did you come to a conclusion about this? Are adj pvalues different when you set the cutoff to 1?

@mt1022
Copy link
Author

mt1022 commented Jul 18, 2024

We didn't try this. Combined and adjusted p values are calculated in this script. The code is hard for me to follow.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants