About nominal p-values

[wososa]

Hi @a-slide @tleonardi ,

Are all the p-values (i.e., GMM_logit_pvalue, KS_dwell_pvalue, KS_intensity_pvalue) adjusted p-values? Is GMM_logit_pvalue equal to GMM_anova_pvalue? Since adjusted p-values will be influenced by the number of genes/transcripts/p-values in a given scenario, is there a way to obtain nominal p-values?

Thanks, Woody

[tleonardi]

Hi @wososa, apologies for the late reply. Yes, all p-values in the results text file are adjusted. The GMM_logit_pvalue and GMM_anova_pvalue are not the same thing. You can find details of the two different tests in the Methods of Nanocompore's preprint on bioRxiv but the recommendation is to stick to the logit test (the Anova test is largely underpowered for normal experimental designs with a small number of replicates). The nominal p-values are stored inside the SampComp.db, which is a normal python pickle. You can use the code for the function that does p-value correction as a starting point to extract uncorrected p-values.