Diversity algorithms

Once a database of candidate molecules has been prepared, it may be desirable to select a diverse set of molecules. Diversity algorithms are designed to select sets of molecules in such a way that the chemical space from which they have been extracted is sampled democratically.1291 Molecules are represented in this space using molecular descriptors and dissimilarity between them is quantified using metrics derived from the value of the descriptors. In terms of descriptors that have been used for fragment molecules,... 

To circumvent the quadratic time complexity that plague many diversity algorithms, Willett and co-workers proposed an alternative approach based on the cosine coefficient of similarity. In this appoach, the diversity of a set of compounds. A, be defined by the mean pairwise intermolecular dissimilarity ... 

The numerical evaluation of integrals is a much studied subject of numerical methods and many diverse algorithms have been developed for such problems. The general mathematical statement is given flie integral ... 

Waldman M, Li H, Hassan M. Novel algorithms for the optimization of molecular diversity of combinatorial libraries. / Mo/ Graph Model 2000 18 412-26. 

Agrafiotis DK. Stochastic algorithms for maximising molecular diversity. J Chem Inf Comput Sci 1997 37 841-51. 

One early step in the workflow of the medicinal chemist is to computationally search for similar compounds to known actives that are either available in internal inventory or commercially available somewhere in the world, that is, to perform similarity and substructure searches on the worldwide databases of available compounds. It is in the interest of all drug discovery programs to develop a formal process to search for such compounds and place them into the bioassays for both lead generation and analog-based lead optimization. To this end, various similarity search algorithms (both 2D and 3D) should be implemented and delivered directly to the medicinal chemist. These algorithms often prove complementary to each other in terms of the chemical diversity of the resulted compounds [8]. 

In this chapter, we provide an overview of selected advances in computational algorithms for the rational selection of molecule libraries for synthesis. Specifically, the following conceptually and algorithmically diverse topics are addressed ... 

QSAR modeling. Therefore considerably larger and more consistent data sets for each enzyme will be required in future to increase the predictive scope of such models. The evaluation of any rule-based metabolite software with a diverse array of molecules will indicate that it is possible to generate many more metabolites than have been identified in the literature for the respective molecules to date, which could also reflect the sensitivity of analytical methods at the time of publishing the data. In such cases, efficient machine learning algorithms will be necessary to indicate which of the metabolites are relevant and will be likely to be observed under the given experimental conditions. 

In conclusion, it is likely that computational approaches for metabolism prediction will continue to be developed and integrated with other algorithms for pharmaceutical research and development, which may in turn ultimately aid in their more widespread use in both industry and academia. Such models may already be having some impact when integrated with bioanalytical approaches to narrow the search for possible metabolites that are experimentally observed. Software that can be updated by the user as new metabolism information becomes available would also be of further potential value. The held of metabolism prediction has therefore advanced rapidly over the past decade, and it will be important to maintain this momentum in the future as the hndings from crystal structures for many discrete metabolic enzymes are integrated with the diverse types of computational models already derived. 

Agrafiotis DK. A constant time algorithm for estimating the diversity of large chemical libraries. / Chem Inf Comput Sci 2001 41 159-67. 

From the diverse possible regression calculi, a certain algorithm has to be selected, namely that of the regression of y onto x see Eq. (6.6). 

On the other hand, the algorithm requires a degree of diversity in the population to make continued progress. If every string in the population is identical, one-point crossover has no effect and any evolutionary progress... 

The basic FEP algorithm for ligand binding can be improved in several ways. One method is to use a nonphysical ligand that serves as the common reference state for a variety of ligands of interest [25]. This method, referred to as the one-step perturbation approach, appears to be quite successful even for complex and fairly diverse ligands [26],... 

---