Systematic Analysis and Prediction of the Target Space of Bioactive Food Compounds: Filling the Chemobiological Gaps

Food compounds and their molecular interactions are crucial
for
health and provide new chemotypes and targets for drug and nutraceutic
design. Here, we retrieve and analyze the complete set of published
interactions of food compounds with human proteins using the FooDB
as a compound set and ChEMBL as a source of interactions. The data
are analyzed in terms of 19 target classes and 19 compound classes,
showing a small fraction of target assignment for the compounds (1.6%)
and unraveling multiple gaps in the chemobiological space for these
molecules. By using well-established cheminformatic approaches [similarity
ensemble approach (SEA) combined with the maximum Tanimoto coefficient
to the nearest bioactive, “SEA + TC”], we achieve a
much enhanced target assignment (64.2%), filling many of the gaps
with target hypothesis for fast focused testing. By publishing these
data sets and analyses, we expect to provide a set of resources to
speed up the full clarification of the chemobiological space of food
compounds, opening new opportunities for drug and nutraceutic design.