EnhancedVolcano R pdf export size

Question

[rbronste]

Hi,

I was wondering if anyone knew a way to make the export of pdfs from EnhancedVolcano significantly smaller in terms of file size for plots containing many points (over 100K). Right now they are like 4mb after externally reducing the size of the pdf before manipulating it in Illustrator which is still too large, starting size is like 20mb.

Wondering if there are any tricks out there, thanks.

Answer 1

[Kevin Blighe]

Reducing the size of the points that are output via pointSize is one way, in conjunction with decreasing the length and width of the output file. You can also reduce the value of pointsize passed to the pdf() function itself.

Another option would be to simply remove variables from the input table that are not statistically significant and that would otherwise never be able to be seen in the plot space. For example, you could identify all variables with |log2FC| < 0.5 and p>0.9, and then sample 50% or 25% of these, removing the remainder. The sample() or seq() functions could be used, in this regard.

Kevin

[rbronste]

Thanks, very helpful! So my input table in this instance is a DESeqDataSet, would you still use sample() or seq() in that case? I have about 325K variables and would def be able to remove 50% within certain FC and pval constraints, probably FC < 2 and pval > .5

[Kevin Blighe]

Well, I mean to reduce the size of just the results table object, and to also clearly report this in the methods, i.e., 'x number of non-statistically significant variables were removed for the purpose of improving visualisation'

Other filtering could become before you even run DESeq2, i.e., for low count variables (?)