When starting a new pharmaceutical chemistry project, there are several common approaches. In my opinion, the most interesting is to take a natural product with known therapeutic properties and begin modifying its structure. This strategy has yielded life saving, world changing medicines including the penicillin based family of antibiotics. Another strategy is to simply perform a high throughput assay with thousands of compounds in vitro. This “brute force” strategy isn’t so reliable in that molecules that may disrupt the assay may be false positives and molecules which require long incubation times may be false negatives. Thirdly, there are computational methods for hit identification. This is where things get interesting.
Finding a hit compound from molecular modelling can proceed through multiple methods. The most interesting to me is using ligand based approaches to identify molecules similar to a known ligand, but still atomically unique to that known ligand. Features like tanimoto coefficients help quantify molecular similarity and make ligand based virtual screening very interesting. Once a library of small molecules is generated, one could theoretically go straight to biological testing. Another option is to take that “enriched” library and use structure based drug design to see which of the “enriched” batch of molecules will bind best to the target structure. Alternatively, one could simply skip the ligand based screening and go straight into “brute force” molecular docking of millions of compounds to see which fit optimally into a structure. These different approaches are what make computational chemistry a field which is fundamentally interdisciplinary and thoroughly philosophical.
