Entering edit mode
3.3 years ago
TheCatalyst
•
0
Can you please help me figure out how the adjusted p-values were obtained in the attached figure. Did they perform Wilcoxon rank sum test and then divided the resultant pvalue with the total number of genes tested?
The figure legend: Expression fold change between cancer (C42B) and normal (RWPE1) cells of genes that are located in large size normal-specific TADs (red, see the symbol ❖ to the right side of graph in c) and not large size normal-specific TADs (blue, solid diamond ♦) (Wilcoxon rank sum test, * adj. p value < 0.05, ** adj. p value < 0.01, *** adj. p value < 0.001)
One cannot deduce the exact method from that partial plot without context (what is a TAD?), I suppose reading the original article and focus on the methods section. Also, the *** seem to indicate highly significant difference, however when looking at the box-plot (which in itself is not ideal) I don't really buy that is something else than an artifact, but it might be due to large number of data points in wilcox test.
Yes, there are hundreds of genes (data points) covered by TADs (Topologically Associated Domains). Ref. Rhie et al., 2019, Nat Commun. (link to Methods section) The methods section doesn't mention the stats analysis.
It does mention it in a way:
And in the Results section referencing Fig.1d:
However, this in context, mentioning only a p-value (not adjusted p-value), together with a single comparison of two box-plots shown in Fig. 1d confirms my interpretation that this is just a single Wilcox test carried out. Thus, the mentioning of adjusted p-value is redundant for this part of the figure, but not wrong, for a single test p-value = adj. p-value. It might serve a purpose for other parts of the panel figure where multiple testing does occur.
Please supply the reference (why supply a figure without a full reference? ;) FWIW, TAD = Topologically Associated Domain...chunks of DNA with physical proximity from HiC experiments.
Please see the response above for the reference.