The impression I am getting is that they have to order the GCMS file for the chems they want to test for.
It is more complicated than that. Good analysis techniques employ two stages:
a) Separation (e.g.: GC or TLC or HPLC)
b) Identification (e.g.: MS, CAD, IR absorbance, UV fluorescence, Stokes/AntiStokes emission).
So in the step
a the sample is separated into different components and later in step
b these components are identified one-by-one.
(Note:
the time of elution or distance of separation is/can be used for identification, too....or aid in it).
Without good separation, you get overlapping identification spectra that are hard to correlate automatically. They look like this:
Note, that certain detectors used for identification (like the one that made the graph above) can "see" the bonds of intact molecules, while some of them, like e.g. MS can only infer them because by the time molecule gets to the detector they are mostly fragmented and their bonds are broken, so it is an educated guesswork how these fragments were bound originally.
Also, most of the time the sample is prepared somehow by dissolving it in a solvent and trituration/filtering or reacting with a derivatizing agent or both. This preparation affects the results of
a &
b.
The results from
a and
b can be correlated automatically be a computer, when the machine is calibrated well but it still the job of a human to spot poor spectra correlation (erroneus library match), overlapping spectra or poor calibration.
The authors of the paper that Hilopsilo quoted, did just that. They spotted closely overlapping peaks on their chromatograms.
Other analysts might use different separation techniques, that are not able to resolve closely related regioisomers or identification methods such as Mass Spectrometry which can easily confuse regioisomers when the relative abundances of their fragments are not accounted for diligently ...or just happen to be identical.
It is not a coincidence that we prefer to use HPLC with Raman and IR absorption detectors at work rather than GC/MS ...and when due diligence is required we even try several HPLC columns for separation ...just to be sure. You'd be surprised, that even with multiple columns sometimes certain components elute at the same time (do not separate well) and only after reacting the sample with derivatizing agents (e.g. PFPA ), a well defined separation occurs.