Until now, the answers to this question have recommended the conclusion of graphs in vector formats. This will give you a better result, allowing you to resize your image as you need for any medium on which your image will be displayed (whether it be a web page, document or presentation) , but this is at the expense of computational cost.
For my own work, I often find it much more convenient to save my stories in a raster format with sufficient resolution. You probably want to do this when your data takes a non-trivial amount of time to build.
Some examples of where I find the raster format are more convenient:
- Manhattan storylines: A plot showing the p-value for hundreds of thousands of millions of DNA markers in the genome.
- Large heat maps: Clustering the top 5,000 differentially expressed genes between two groups of people, one with the disease and one healthy.
- Network rendering: When drawing a large number of nodes connected to each other along the edges, redrawing the edges (in the form of vectors) can slow down your computer.
Ultimately, it comes down to a compromise in your own sanity. What annoys you more? Does your computer grind while trying to redraw the image? or find out the exact dimensions to make the image in raster format so that it does not look terrible for your ultimate publishing tool?
source share