Combinatorial Library Design:  Maximizing Model-Fitting Compounds within Matrix Synthesis Constraints

Abstract

The use of combinatorial chemistry has become commonplace within the pharmaceutical industry. Less widespread but gaining in popularity is the derivation of activity models from the high-throughput assays of these libraries. Such models are then used as filters during the design of refined daughter libraries. The design of these second generation libraries, which efficiently test and conform to the derived activity model from the large space of virtual possibilities, remains an area of considerable research. We present here a computationally efficient method for the design of optimally dense (in model matching compounds) synthetic matrices from in silico virtual libraries.

Focused Library Design Based on Hit and Target Structures: Method and Application in Drug Discovery

“…D max and D min are the maximum and minimum D k values of a specific generation, respectively. It has been estimated that $40% of compounds fail to be developed into drugs due to their poor pharmacokinetic properties (Stanton et al, 2000). The reasonability of the descriptor set for calculating molecular diversity was verified by two libraries, No.…”

Multiobjective optimization of combinatorial libraries

“…While for simple cost functions several very effective greedy algorithms can be employed [41,42], arbitrary objective functions have unpredictable surfaces with many local minima, and require a stochastic approach that is in principle suitable for identifying global minima. While for simple cost functions several very effective greedy algorithms can be employed [41,42], arbitrary objective functions have unpredictable surfaces with many local minima, and require a stochastic approach that is in principle suitable for identifying global minima.…”

“…From an algorithmic perspective, the most notable examples include the independently developed simulated annealing implementations of Agrafiotis [9 -12] and Hassan et al [13,14], and their subsequent variations by Good and Lewis [15] and Zheng et al [16]; Brown and Martin's genetic scheme to generate libraries designed to minimize the effort required to deconvolute biological hits by massspectroscopic techniques [17]; the attempts by Gillet et al [18], Rassokhin and Agrafiotis [19], and Brown et al [20] to enforce certain property distributions on the final design; and the latest use by Sheridan et al [21] of genetic algorithms for designing targeted libraries. They represent only a small fraction of proposed library design methodologies, which range from conventional experimental design [4,30,31] to clustering [32] and cluster sampling [33], conformational sampling [34], partitioning [35,36], Boolean logic [37][38][39], vector analysis [40], and some more recent "greedy" algorithms for selecting combinatorial arrays [41,42]. They represent only a small fraction of proposed library design methodologies, which range from conventional experimental design [4,30,31] to clustering [32] and cluster sampling [33], conformational sampling [34], partitioning [35,36], Boolean logic [37][38][39], vector analysis [40], and some more recent "greedy" algorithms for selecting combinatorial arrays …”

A new approach to the rationale discovery of polymeric biomaterials

“…In the pharmaceutical industryadvanced tools of drug discovery (combinatorial synthesis, highthroughput experimentation, computational modeling) have revolutionized the way lead compounds are identified [20][21][22][23]. These general concepts of advanced drug discovery are in principle applicable to materials discovery as well.…”