UPPSALA, Sweden—The chemical diversity offered by natural products is only matched by the diversity of biological activities of these compounds and yet most efforts to thoroughly catalogue these compounds has been limited to those with traditional druglike properties. Recognizing that there is an even greater spectrum of possibilities in nature, researchers from Uppsala University and AstraZeneca R&D Mölndal set out to generate a more complete record.
As they describe in the Journal of Natural Products, the researcher mapped the biologically relevant chemical space by defining the rules of chemical characteristics, which they compare to lines of latitude and longitude, and key core structures, which they liken to cities. They then established the boundaries of this chemical global positioning system (ChemGPS) using satellite compounds that lay outside of the natural product space because of extreme values for one or more desired properties.
Using the Dictionary of Natural Products' 167,000 products as a starting set, they performed cluster analysis more than 10,000 clusters, of which the 1376 clusters comprised of more than 50 compounds were used to model the initial ChemGPS-NP (natural product). The map coordinates for new compounds are t-scores from principal component analysis from 35 molecular descriptors.
The researchers see ChemGPS-NP as providing a framework for improving the efficiency and stringency of compound comparison and selection, letting drug developers identify volumes of chemical space central to specific biological activities. Together, they suggest, the benefits of their system increase the probability of hit generation for researchers searching for natural products with novel bioactive properties.