Machine intelligence design of 2019-nCoV drugs

Kaifu Gao; Duc Duy Nguyen; Rui Wang; Guo-Wei Wei

The top-ranking candidate of our generated molecules is MSU3298 (see Figure 5 ). Its predicted binding affinity to the nCoV-2019 protease is -10.56 kcal/mol, which is higher than that of the best candidate CHEMBL222234 (-10.02 kcal/mol). The high binding affinity is due to the existence of many hydrogen bond acceptors and donors and forming a strong hydrogen bond network with nCoV-2019 protease. For example, the strongest hydrogen affinity is also due to the strong interaction at the tail of the molecule. One Cl atom on the quaternary ring forms a hydrogen bond with the side chain of residue Arg188. One O atom in this tail also has a hydrogen bond with the main chain of residue Tyr54. In the head of the molecules, the two Cl atoms interact with the methanethiol of residue Cys145 and the main-chain amino of residue Glu166. These hydrogen bonds promise a strong binding to 2019-nCoV protease binding site. The third molecule is MSU3245 (see Figure 7 ) with a binding affinity -9.55 kcal/mol. The strongest hydrogen bonds between this molecule and the protease are the Cl atom on the benzene ring of the molecule and the side-chain hydroxyl of the residue Ser144. Additionally, the Cl atom of the ternary ring and the methanethiol in residue Cys144.