CC(=O)NCCc1cccc(CC(=O)Nc2cccnc2)c1
NS(=O)(=O)c1cccc(CC(=O)Nc2cccnc2)c1
COc1cccc(NC(=O)Cc2cccc(CCNC(C)=O)c2)c1
NS(=O)(=O)c1cccc(CCC(=O)Nc2cccnc2)c1
NS(=O)(=O)Cc1cccc(CC(=O)Nc2cccnc2)c1
COc1cccc(NC(=O)Cc2cccc(CS(N)(=O)=O)c2)c1
CC(=O)NCCc1cccc(CC(=O)NCc2cccnc2)c1
Two of the key sites of interaction between many of the ligands is residue E166 and H163. Ligand x0104 is an example the forms a H-bond with the carbonyl of E166 and Ligand x0107 forms a H-bond with the NH of E166 and the NH of H163. The two ligands occupy different binding pockets however. It appears that the two pockets could be bridged by a linker that fits onto a more hydrophobic space. The suggested molecule is one example of what I am proposing by judging the distances and conformational flexibility needed to occupy both pockets and make similar binding interactions. Multiple other ligands also dock into the same two pockets and extend to nearly bridging, but they do not actually extend into both pockets. Other ligands binding similarly to x0104 are x0161, x0874 and x0946. Ligands binding similarly to x0107 are x0397, x0426, x0434, x0678 and x0967. Ligand x0967 offers another way to bind H163 with a -OMe group instead of a pyridyl Nitrogen.
I do not have any docking software available to me to test the ligand ability to bind effectively. I downloaded all of the pub files and examined them in a freeware protein viewer known as MolSoftBrowser. I can share captured images or the MolSoftBrowser files if needed. I did not attach since it appeared that you only wanted pub files uploaded.