Plots for DNA methylation vs Gene Expression

Question

[sswang25]

I have trying to visualize the correlation between DNA methylation and Gene expression in my dataset of 52 unique samples. I have DNA methylation beta values, and gene expression normalised VST values for each sample.

DNA methylation values are in a matrix - with samples as columns, and probes as rows:

Gene expression counts are in a matrix with samples as columns and probes as rows.

I have run a simple correlation between DNA methylation value and Gene expression value. I used all 689,000 probes against 22 genes of interest. I have generated a simple dataframe with the correlation R values

I have also generated a dataframe with the corresponding P values

I would like to visualise these values - but I have no idea what kind of plot to use, and how to go about doing it.

I have read other papers where they have generated starburst plots with TCGAbiolink package, that look like this:

I have tried using the TCGAbiolink package but I cannot make the summarizeexperiment object needed from my current data.

**my questions are:

1. Does any have any suggestions for how to visualize my raw correlation data?
2. How can I make a Starburst plot like the one shown without TCGAbiolink?**

[seidel]

I have no idea what kind of plot to use, and how to go about doing it.

Knowing the plot to use begins with knowing the question you are asking. What, exactly, is your question?

Also, you've calculated correlations, but the starburst plot is based on log transformed P-values for differences in gene expression between two conditions, and differences in methylation between two conditions (where the log transformed, FDR adjusted p-values are multiplied by 1 or -1 depending on the direction of the ratio from the elements being compared (i.e. gene expression higher in one condition than another). This is different than the data you calculated, isn't it?

[sswang25]

Thanks for explaining, The question I am asking is DNA methylation changes in which probes are correlated with gene expression changes in which genes? Considering this, I think that plotting the starburst plot would be much better. However the problem is I am researching rheumatoid arthritis - responders/non-responders to treatment. There isn't a huge amount of difference in methylation and gene expression between the two groups. There are only 43 differentially expressed genes using DEseq2, and no differentially methylated probes. Therefore I was thinking, I could plot the nominal P-value of the DNA methylated probes instead on the starburst plot.

Additionally I am having difficulty using the TCGAbiolink package - as I cannot make a summarizedexperiment object from my data.

Therefore I was wondering how to make a starburst plot myself or if there are other packages that can do this?

Thanks for all your help