WITHIN the color labs of the world's 12 major paint companies, chemists continually tinker with formulas, searching for new effects that will dazzle and entice the auto-buying public.
In the formula search, data from 50 randomly selected samples of total 80 samples were used for model training.
Additional parameters to assist in the identification of chemicals based on molecular formula search hit lists is required, and predicted RTs can be a valuable parameter for this purpose.
Thus, the quality of the molecular formula search results depends greatly on the properties of the database.
Indeed, molecular formula searches of the DI-FT/MS dataset using PubChem afforded results with high completeness (α = 0.72).
Perform a molecular formula search against each of the databases using a suitable threshold value (Δthres = σ, 1.5σ, or 2σ).
These results suggest that PubChem is too large and KNApSAcK and KEGG are too small to perform an accurate molecular formula search of plant metabolome data.
When the analytical error is larger than the threshold value, the molecular formula search will not obtain a correct answer from the database (i.e., a false negative).
Mass analysis accuracy at FT-MS levels (σ = ∼0.5 1 mDa) is likely required to obtain a molecular formula search with a low FDR value (<10%; Table 1).
This suggests that the absolute unit (mDa) rather than the relative unit (ppm) is more suitable for defining threshold values for molecular formula searches.
