Hi Pete, just getting off call. The procedures the UPenn folks used is unknown to me but I will say this: I think it would have been less efficient to blindly add moieties to an aryl-amide backbone and test all these possibilities. The chemical composition and stereochemical structure of the defensins are known and characterized. The structure of some of their ligands, what they attach to have been described, at least some of them, eg in cell wall of bacteria and in leukocytes and epithelial cells. It would make sense to me to start with different side chains which had a higher probability of successful binding given the chemistry of the ligands. But some random computer generated structures could have been used since the complete list of defensin targets are not known.

Also, in response to one poster, although new antibacterials are needed, new antifungals are desperately needed. Because fungi are higher form eukaryotes with a high degree of homology with humans, the specificity of available antifungals to fungal targets is compromised and there is bleed over to human processes which cause side effects. To have an antifungal with limited side effects but high efficacy would be HUGE