Mass error cut-off different when using Pubchem as database

[janajohansson]

Hi! I noticed that the cut-off for mass error is different when using Pubchem compared to using other databases in the web version of Metfrag. For example, if I set "search ppm" to 5 and search for a neutral mass of 183.01606 various databases (Pubchem lite, Norman, Comptox) suggest structures with the elemental formula C7H5NO5 (mass error 3.9). However, if I do this with Pubchem as the selected database this elemental formula is not suggested (unless I increase "search ppm" to 6).

[schymane]

Thanks Jana! I can confirm this observations, the formulas do not appear at 5 ppm but are there again at 5.5 ppm when using the PubChem query from the "server databases" listing in MetFragWeb. The monoisotopic mass in the PubChem candidates is only reported to 3 d.p. in the candidate files, which could explain this. @sneumann is this an issue with the mirrored copy of PubChem?

From PubChem query (ppm 5.5):

From PubChemLite (ppm 5):

The masses in PubChemLite reflect the masses in active records:
https://pubchem.ncbi.nlm.nih.gov/compound/6807#section=Computed-Properties