Hello,

Sorry for asking a lot of questions about clusterProfiler. I try to fully understand the results and graphs I want to use.

When using dotplot() on the result of enrichGO() and enrichDO(), I always observe that the most significant terms are the one with the biggest number of genes and the highest gene ratio. Whereas, we can imagine to get a very significant term with very high gene ratio but small (at least not the biggest) number of genes...

I observed that also on all the examples I saw on the vignette and on the web.

What is the rational behind this? This way, the small gene sets can never appear significant...

Jane

pvalue is not correlated with gene ratio, see the color of the plot here.

Indeed, pvalue is not correlated with gene ratio, in my examples as well. But I did not mention this relashionship. I speak about the positive correlation between the number of genes and the gene ratio. Can you answer this please?

Your example illustrates perfectly my point: one more case with no high gene ratio and small number of genes, as we can expect.

Thank you for your answer. I thought n was the number of genes in a category and not the size of the input list of genes of interest. That is why I did not expect correlation between the number of genes and the gene ratio.

It would be great to have the possibility to plot this information as well (ratio of genes of interest over number of genes in a category), that is adding a third possible value to "x".