Libraries property-based

In general, reagent-based selection is much faster and more convenient to execute in the laboratory as compared with the product-based selection. On the other hand, the latter strategy usually provides more accurate results. There exists a potential to combine both approaches to achieve more optimal results, particularly in the case of large exploratory virtual combinatorial libraries, for which mass random synthesis and screening are not economically feasible. In this article, we demonstrated the usefulness of property-based approach for selection of optimal GPCR ligands. 

S. A., Okun, L, Ivashchenko, A.A., and Savchuk, N.P. Property-based design of GPCR-targeted library./. Chem. Inf. Comput. Sci. 2002, 42, 1332-1342. 

Another use for combinatorial libraries has been the screening of peptides for antimicrobrial properties. In this case, the design of the library is based on antimicrobial peptides found in nature. A combinatorial synthesis is used to find alternative unnatural amino acids expected to mimic the antimicrobial properties.23 Peptide libraries also have been used to find compounds that could bind the lytic peptide mellitin.24 The library was synthesized in solution phase, purified, and evaluated using time-of-flight mass spectrometry (TOF-MS). The sequences determined to bind to mellitin contained hydrophobic pairs. By binding to mellitin, they were able to prevent the cell-surface mellitin interaction. This is an example of a peptide library able to afford compounds that interact with other small peptides without having to find an interacting protein first. 

Our design process calls for ab initio specification of all reaction products for direct use in subsequent molecular selection (diversity) processes. Other approaches base the selection of compounds for synthesis solely on reagent properties. Thus the diversity of the library is based on reagent properties rather than those of the final products. Although this is inherently easier to carry out it can detrimentally... 

The strategy we employed to reduce the size of the virtual library was based on independent analysis of the properties of individual building blocks, in order to select subsets of suitable and diverse reactants from a pool of available chemicals. 

Perry, N Selection of diverse database subsets by fingerprint and property-based methods. Paper presented at the Chemical Structure Association/Molecular Graphics and Modelling Conference on Computational Approaches to the Design and Analysis of Combinatorial Libraries , University of Sheffield. 14-16 April 1998. 

Figure 13.2 Multiple property-based library shaping. Property distributions are shown for the raw collection (gray bars) and for the selected library members (black bars). The drug-likeness score corresponds to the output value of an artificial neural network, where a value of 1 indicates maximal drug-likeness (for details, see the text). The rule-of-5 violations are counted per molecule.

References for articles reporting results of ab initio quantum chemical calculations on different properties of aniline and derivatives can be retrieved from the compilation Quantum Chemistry Library Data Base (QCLDB), Japan Association for International Chemical Information, Tokyo, Japan, 2005. 

D and 3D autocorrelation vectors [70] represent intramolecular 2D topologies or 3D distances within particular molecules. An autocorrelation coefficient is a sum over all atom pair properties separated by a predefined number of bonds (2D) or distance (3D), while the entire vector represents a series of coefficients for all topological or Cartesian distances. Atomic properties involve hydrophobicity [71], partial atomic charges, hydrogen bonding potential and others. Again, a PCA is often used to reduce the number of variables. 3D autocorrelation vectors of properties based on distances calculated from 3D molecular surfaces [72] have also been applied to visually assess the diversity of different libraries [73]. 

It is relatively easy to include additional library properties such as druglike physicochemical property profiles in the product-based approaches to combinatorial subset selection. For example, several groups have handled multiobjectives through the use of weighted-sum fitness functions. In the SELECT program [89], multiple objectives are handled via a fitness function such as the one shown below ... 

An analysis of compounds in the World Drag Index has resulted in the rale-of-five, which states that poor oral absorption is expected for compounds with a molecular weight above 500, a calculated octanol/water partition coefficient >5, the number of hydrogen bond donors >5, and the number of hydrogen bond acceptors >10. Exceptions to this rale may be compounds with an active uptake mechanism. Such simple rales are now widely used in the design of combinatorial libraries. More detailed knowledge on relevant properties can be used to optimize oral absorption and pharmacokinetics, an approach which has been called property-based design. ... 

The most important results and facts, in particular the extensions of the MOLG EN class library Reaction-based generation of structures, QSPR studies using different kinds of molecular descriptors, various methods for prediction, ranking and classification of mass spectra, relations between spectra and properties and CASE using electron impact (El) mass spectrometry. 

---