The loadGSC function is used to parse a file describing all gene to gene-set associations into the format used by downstream piano functions (like runGSA). In its simplest form it can be a two column text file with gene-sets in one column and genes in the other, and accordingly one gene to gene-set association per line. As you mention it is also possible to use the gmt-format provided by the MSigDB.
It is essential that the gene IDs of your data are the same as the ones used in the gene-set collection (i.e. the file you use as input to loadGSC). If this is not the case, you need to translate one of your set of IDs to match the other. For this you can e.g. use the biomaRt package or the different Bioconductor Annotation Packages (I am not sure where your data comes from but as an example both the yeast2.db package and the org.Sc.sgd.db package provide mappings between various gene IDs).
You don't mention what kind of gene-sets you are interested in analyzing, but note that you can use the packages mentioned above to construct a GO-term gene-set collection, without using the ones from MSigDB (which mainly use, if I am not misinformed, human gene IDs).
I hope this helps!