ChemGPS-NP and the Exploration of Biologically Relevant Chemical Space

Sammanfattning: Chemical space is basically infinite, and comprises all molecules that could possibly exist. Intelligent ways to efficiently navigate through chemical space and to select promising compounds in drug discovery are important tasks, and the focus of this thesis.In this work a new model for chemical space navigation, ChemGPS-NP, was developed. This model is based on a methodology where a global chemical space map is defined through principal component analysis of physico-chemical properties of a reference set of compounds. Through interpolation from the reference set, positions of novel compounds can be defined on this map and interpreted as chemical properties.ChemGPS-NP was demonstrated to be able to chart the entire biologically relevant chemical space, including both drug-like and natural compounds. This is an important improvement considering the present interest in natural products (NPs) in the pharmaceutical industry, as well as the track record of NPs to serve as basis for more than 50% of all marketed drugs. ChemGPS-NP proved able to handle and process large data sets, to aid in efficient selection of test objects, and to extract useful information from the results of high-throughput screening campaigns. Using ChemGPS-NP, it was shown that NPs occupy unique regions of chemical property space in comparison to drug-like compounds, and a number of features distinguishing NPs from medicinal chemistry compounds were identified.ChemGPS-NP was also shown to be able to reliably predict mode of action of anticancer agents based on chemical structure, a finding that has potential to improve cancer research efficiency. Applying a property based similarity search based on calculated eight dimensional Euclidean distances from ChemGPS-NP rendered a tool to identify NP inspired potential leads for drug discovery.Furthermore, ChemGPS-NPWeb, an online version of ChemGPS-NP, was developed, which provides scientists with open access to the tool via http://chemgps.bmc.uu.se/.