Some peak with mass m and a molecular formula with mass (m') match if (left|m^{prime }-mright| le MA cdot m).
As a baseline method to evaluate against, we use the naïve method that returns the molecular formula with the smallest mass difference to the measured parent mass.
These spectroscopic features were very similar to those of cudraisoflavone L, which was recently isolated from the same genus and shared the same molecular formula with 4 [26].
We compare the (undisturbed) simulated isotope distribution of the true molecular formula with the simulated isotope distribution of each candidate molecular formula, by summing the absolute errors of intensities over all peaks.
But both intrinsically charged molecules and free radicals are comparatively rare; to this end, we use (P_{text{rdbe-odd}} = 0.1) for molecular formulas with even sum of valences, and (P_{text{rdbe-odd}} = 1) for all others.
We identify the molecular formula of a compound as proposed in [34]: for each molecular formula decomposition that is within the given mass accuracy of the parent peak, find the FT rooted at this molecular formula with maximum score (maximum posterior probability).
This formulation of the problem is not easily accessible by algorithmic means; to this end, we use an alternative formulation which, for additive scorings, is equivalent to the above [34]: for each peak in the fragmentation spectrum, we find all molecular formulas with mass difference sufficiently small.
These approaches are unfavorable in theoretical complexity as well as in practice: for the elements CHNOPS there exist more than 7 × 10 molecular formulas with mass ≤1500 Da.
The "Find" function in Microsoft Office Excel was used to match the empirical molecular formula with that of published compounds within the library.
