These steps are largely independent; any one may be replaced with a different function, with some work. For example, the default clustering is according to Gene Ontology terms, but another clustering is also fine. Similarly, although the package is designed for mass spec experiments, it could in principle be used with other sorts of data, provided that it is based on Uniprot IDs, corresponding protein symbols (e.g. CTCF), and some sort of score.
The output format is svg (scalable vector graphics), suitable for editing with Adobe Illustrator or similar.
Please be aware that this program does not produce finished, polished figures. They usually need some editing to look right. Also, a few iterations are usually needed to decide exactly which GO terms are most meaningful (at the step where you manually select the set of interest).
Caveat: These plots mis-represent the clustering into GO categories. Most proteins are in more than one category; most GO categories have more than one parent. These plots do not reflect this, but instead show any given protein in only one category (the one with fewest members in this plot), and any given GO category under only one parent (the one with fewest...).