Combi-Chem

4i. Combinatorial design

Introduction

The combinatorial design approach found in the C²·LibSelect module combines advantages from reagent selection and product selection. It provides the best of both worlds, since it considers diversity in product space but operates on selections of reagents. Product-based reagent selection (also termed R-group subsetting) provides selections of reagents that optimize the diversity of the products.

The combinatorial design capabilities are accessed from the Rgroup Subsetting menu items in the LIBRARY ANALYSIS card in the COMBI-CHEM I deck:

• The Rgroup Subsetting/Diverse Library menu item enables you to select sets of fragments in R groups so as to generate a combinatorial library with maximum diversity.

• The Rgroup Subsetting/Focused Library menu item enables you to select sets of fragments in R groups so as to generate a combinatorial library that focuses around the properties of one or more target models.

Note

Diverse combinatorial design

Subsets of fragments from each R group are selected so as to generate a combinatorial library with maximum diversity. You specify the number of fragments to select for each R group. Then a stochastic optimization procedure is used to optimize the diversity of the corresponding library, based on the properties of the whole molecules.

Three diversity metrics for R group subsetting are available:

• Cell-based Fraction: Allows for the selection of R group fragments so as to optimize the number of occupied cells in the resulting library.

• Cell-based Chi2: Selects R-group fragments so as to optimize the chi-squared distribution of models in all the cells.

• Max MinSpanTree: Used for both R group subsetting and selection of diverse models from the study table. This function is based on calculating the minimum spanning tree for the subset of points to be selected, which is then used to compute an error function based on the length of each edge in the tree.

You can specify fragments in R-groups that should always be selected when doing an Rgroup subsetting optimization (combinatorial selection) to design a diverse library, a focused library, or a restrained library. Fragments are fixed using the Fix Fragments control panel, which is opened by selecting the Fix Fragments in Rgroups pushbutton in any of the Rgroup subsetting control panels.

Fix and unfix fragments by clicking the fix cells to toggle the Yes/No "fix state" of each fragment.

There is no need for an exact one-to-one correspondence between fragment combinations and products in order to apply the Rgroup subsetting method. You can have datasets with fewer or more products than the total number of R-group fragment combinations.

Note

Focused combinatorial design

Select subsets of fragments from each R group so as to generate a combinatorial library that focuses around one or more target models. You specify the number of fragments to select for each R group. Then, a stochastic optimization procedure is used to focus the corresponding library around the target models, based on the properties of the whole molecules and the properties of the target molecules.

The metric used to focus a library is based on the distances in property space between the target models and the models in the library. The metric is defined as a combination (with user-defined weights) of the minimum distance and the average distance between target models and library models.

Note