Herein is described a method to rapidly screen a large chemical space for a compound that binds to a target protein through an iterative fragment assembly approach that can be performed at low reagent cost and without requiring purification of the assembled product. The method employs a library of test ligands each of which comprise a 'bait' molecule, which is known from prior art or prior screening to have some intrinsic affinity for the target protein, and a test moiety.