In order to build an interaction network you need information regarding protein interactions, gene interactions, etc. This information only exists at genome scale for some selected model species. If you would like to use A. thaliana which presumably is the closest model species (closer than animals, at least) you need gene annotation. The programme won't know what's interacting with what without any gene names. The easiest thing to do is to annotate your organism based on A. thaliana (or a close species if is sequenced and annotated) and keep in mind the bias you're introducing. Then generate the interaction network using A. thaliana as interaction database. Good luck.