文摘
A genetic algorithms (GA) based strategy is described for theidentification or optimization of activeleads. This approach does not require the synthesis and evaluationof huge libraries. Instead it involves iterativegenerations of smaller sample sets, which are assayed, and the"experimentally" determined biological response isused as an input for GA to rapidly find better leads. The GAdescribed here has been applied to the identificationof potent and selective stromelysin substrates from acombinatorial-based population of 206 or 64 000 000possiblehexapeptides. Using GA, we have synthesized less then 300 uniqueimmobilized peptides in a total of five generationsto achieve this end. The results show that each successivegeneration provided better and unique substrates. Anadditional strategy of utilizing the knowledge gained in eachgeneration in a spin-off SAR activity is described here.Sequences from the first generations were evaluated forstromelysin and collagenase activity to identifystromelysin-selective substrates. GlyProSerThr-TyrThr with Tyr as theP1' residue is such an example. A number ofpeptidesreplacing Tyr with unusual monomers were synthesized and evaluated asstromelysin substrates. This led to theidentification of Ser(OBn) as the best and most selectiveP1' residue for stromelysin.