In Silico Fragment-Based Generation of Drug-Like Compounds

摘要

During virtual library construction, the ability to focus the potential combinatorial explosion of generated molecules on a desired region of chemical space is paramount. As such, de novo molecule generating programs must strike a balance between the freedom to explore new chemical space and the limitations that must be imposed on growth in order to achieve desired features in the generated compounds, such as stability in water, synthetic accessibility, or drug-likeness. With this in mind, the Fragment Optimized Growth (FOG) algorithm was developed to statistically bias the growth of molecules with desired features. At the heart of the algorithm is a Markov Chain which adds fragments to the nascent molecule in a biased manner, depending on the frequency of specific fragment -fragment connections in the database of chemicals on which it was trained. We demonstrate that FOG generates synthetically feasible compounds, and that it can be trained to grow new molecules that resemble desired classes of molecules such as drugs, natural products, and diversity-oriented synthetic products. In addition to generating virtual libraries of compounds, FOG is well suited to expand experimental fragment hits during lead optimization.