The chemically inert noble gases display a surprisingly rich spectrum of useful biological properties. Relatively little is known about the molecular mechanisms behind these effects. It is clearly not feasible to conduct large numbers of pharmacological experiments on noble gases to identify activity. Computational studies of the binding of noble gases and proteins can address this paucity of information and provide insight into mechanisms of action. We used bespoke computational grid calculations to predict the positions of energy minima in the interactions of noble gases with diverse proteins. The method was validated by quantifying how well simulations could predict binding positions in 131 diverse protein X-ray structures containing 399 Xe and Kr atoms. We found excellent agreement between calculated and experimental binding positions of noble gases. 94 % of all crystallographic xenon atoms were within 1 Xe van der Waals (vdW) diameter of a predicted binding site, and 97 % lay within 2 vdW diameters. 100 % of crystallographic krypton atoms were within 1 Kr vdW diameter of a predicted binding site. We showed the feasibility of large-scale computational screening of all ≈60 000 unique structures in the Protein Data Bank. This will elucidate biochemical mechanisms by which these novel 'atomic drugs' elicit their valuable biochemical properties and identify new medical uses.