(A) Composition of the assembled CCS database for drugs and drug
metabolites (dmCCS). Metabolic modification abbreviations: GSH, glutathione
conjugated metabolites; Glc, glucuronide metabolites; Ox, oxidative
metabolites (e.g., −2H, +O, −Me). (B) PCA projections
of the dmCCS database (color) from a PCA computed using the CCSbase
database (gray), colored by CCS. (C) PCA projections of parent compounds
(blue) and metabolites (red) from the dmCCS database from a PCA computed
using the CCSbase database (gray). (D) CCS vs m/z of parent compounds (blue) and metabolites (red) from
the dmCCS database overlaid on the CCSbase database (gray). Dotted
lines represent individual power fits for parent (chartreuse) and
metabolite (orange) data, and residual CCS from these fits are included
below the main plot. (E) PCA projections of dmCCS database computed
using MQNs as molecular descriptors, colored by CCS. (F) PCA projections
of dmCCS database computed using MD3Ds as molecular descriptors, colored
by CCS. (G) PCA projections of dmCCS database computed using the combination
of MQNs and MD3Ds as molecular descriptors, colored by CCS. (H) Individual
feature loadings for principal component 1 from PCA computed on dmCCS
using MQNs as molecular descriptors. (I) Individual feature loadings
for principal component 1 from PCA computed on dmCCS using MD3Ds as
molecular descriptors. (J) Individual feature loadings for principal
component 1 from PCA computed on dmCCS using the combination of MQNs
and MD3Ds as molecular descriptors.