Thanks, arcniz --
Cat in blender analogy is a good one! So good, in fact, that I'll hijack it a bit -- what if we do in fact have 14 specimens of FelineFrosty™ aka cat-in-a-blender, and four were terrific mousers, nine were so-so, and one was an unabashed mouse-sympathizer.
While there is no guarantee that an analysis of the spectral signatures would establish a high correlation of any feature with mouse-aggression, IF (a speculative if) such a relationship did emerge, wither with or without a pre-existing hypothesis to explain it, surely that is probative, and at least could indicate a direction for future inquiry!
BTW -- T. gondii could account for odd cat-mouse interaction behavior if it were the mice we were looking at, but would likely not show up using such a crude method as GC/MS, so your point is somewhat reinforced in my mind -- physical tests that demonstrate properties important to the material's suitability as a fuel while in the conditions observed, are probably the best hope.
Thanks again!