Filtering P-values before comparison

[Jwong684]

Hi CytoSig developers,

From my understanding of the output files, output_prefix.Zscore is in essence the activity level and output_prefix.Pvalue represents how significant the cytokine activity is in a given cell relative to 0. Positive Z-score values would indicate that the activity is highly active in your cell and negative Z-score values would indicate that an activity is repressed in your cell - is that an accurate interpretation?

In another thread, you mentioned that you should filter cells for FDR < 0.05 before comparison between cell types. What about in the context of the same cell type between different conditions? (i.e. if I want to compare the signal activity of malignant cells in subgroup A vs. malignant cells in subgroup B, would I first filter out significant cells first, or just perform a wilcox test of the Zscore between the two groups?)

Thanks for your help!

[data2intelligence]

Hi, Jwong684,
Yes, your interpretations are all correct.

For your second question, please do not filter out any cells if you want group comparisons. Just include all Z-scores and run wilcoxon rank-sum test of all z-scores between two groups.

Sincerely, Peng

[Jwong684]

Thank you Peng - that was very informative.

Just as a follow-up, would you interpret negative Z-score values here as simply, for example "there is no evidence of TGFB1 acting on this cell of interest" or rather, "there is evidence suggesting that the TGFB1-signaling pathway is actively being repressed in this cell of interest"?