Which spectrum should be used identification without an internal library, and how is the “representative spectrum” generated?

[voledtalies]

I am using MS-DIAL (4.9.221218) for GC-EI data processing without any internal spectral library. I found an alignment spot that appears only in one biological class (three replicate samples) and I would like to identify this compound using MassBank.

In the Alignment Spot Viewer, MS-DIAL shows a representative EI spectrum, while in the Peak Spot Viewer each replicate sample has its own deconvoluted EI spectrum. I am unsure which spectrum is most appropriate to export and use for a MassBank similarity search:

the representative spectrum from the Alignment Spot Viewer, or

the deconvoluted EI spectrum from one selected replicate sample?

Additionally, it is not transparent to me how exactly the representative spectrum is generated. Is it simply taken from one specific sample (e.g., the most intense or highest-quality replicate), or is it some kind of merged/averaged spectrum across samples?

Thank you :)