ADMET Predictor® User Manual

4.4.1. PAIRS Menu

4.4.1. PAIRS Menu

Find Similar

Used to find pairs with similar transformations within the spreadsheet.

Generate Clusters

All the pairs can be assigned to clusters in which each member pair exhibits the identical structural transformation. Details of this feature are found in Clustering Matched Molecular Pairs.

Rank Clusters

Orders the clusters such that those at the top represent the largest average change in a user-defined property; the property is selected on an options dialog.

Delete Clusters

Deletes the generated pair clusters.

Generate Transforms

Used to export selected pairs as transform rules. A detailed explanation can be found in Generating Reaction Transforms from Pairs.

Sort By Significance

Sorts the pairs by significance, where significance is the difference in property value divided by structural dissimilarity, calculated as one minus the Tanimoto similarity.

Difference Query

Used to search for particular structural changes between the two molecules in each pair. For example, it is possible to search for pairs in which the second molecule has a benzene ring that is absent in the first molecule. To perform this operation, you would choose the Right Compound menu option, and enter a query for benzene. In order to specify that the extra benzene ring is attached directly to an atom shared by both molecules rather than an atom unique to the second molecule, it is necessary to include an attachment, or “scaffold” atom in the query. ADMET Predictor supports a new SMARTS atom primitive [~s] for this purpose; thus the query [~s]c1ccccc1 specifies a benzene ring attached directly to the common scaffold of the two molecules of a pair. The analogous SMARTS primitive for scaffold bonds is “+”; hence the query [~s]+O specifies an unmatched Oxygen attached to a common scaffold atom through a common bond. Note that SMARTS atoms and bonds not labeled with [~s] or + will match both scaffold and non-scaffold atoms; you can use the “not” operator (e.g., [!~s]) to force matching to only non-scaffold atoms and bonds.

Transform Query

Used to search for pairs exhibiting a specific structural transformation. Queries must be based on Daylight SMIRKS syntax. For example, the query [c:1]>>[c:1][*] will search for any atom being added to an aromatic carbon atom, while [*:1]>>[*:1][OH] will search for a hydroxyl group being added to any atom.

Change Property

Usually the generation of pairs involves the selection of a single molecule property, such that a pair of molecules is only retained if the absolute value of the difference in value of that property for the two molecules is greater than or equal to a user-defined threshold. Both the individual values and the change in value of this property are displayed by default in the Pairs tab. You can change the property to a different one by choosing this menu item. This action will change the values in the spreadsheet.

Define Context Scaffold

The Mismatches column can be used to sort the list of pairs by the number of atom and bond mismatches occurring between the two molecules in each pair. While this is useful for finding pairs exhibiting the smallest structural change, sometimes a you might be interested in finding the pairs with the fewest mismatches occurring within a particular substructure. This menu item can be used to lasso a particular substructure, after which the pairs in the spreadsheet will be re-ordered such that pairs at the top have the fewest mismatches in the specified region, and a new Scaffold Mismatches column will be displayed. Note that this option can only be used when all of the pairs in the spreadsheet include a single reference compound.

Compound Order

If you select a compound property when pairs are first generated (see below), each pair of molecules in the final pair set will be ordered such that the molecule on the left has the lower value of the specified property and the molecule on the right has the higher value. If no property is selected initially, the ordering will be such that the smaller molecule (defined as atom count plus bond count) is on the left and the larger molecule is on the right. The order of selected molecule pairs can be changed using this menu option. Note that the structural depictions themselves will not change when the molecule order is changed. You can change the compound order based on six options which include increasing or decreasing values of size, property and name.

Compound Alignment

By default, the two compounds in a pair are aligned by restricting the scaffold (shared) atoms of one of the compounds such that they have the identical 2D coordinates as the analogous atoms in the other compound. The remaining non-scaffold atoms are then re-derived using a special depiction algorithm. In some cases this protocol can lead to significant distortion, particularly when the common atoms represent partial ring structures. When significant distortion occurs, you may wish to align the compounds using simple rotation rather than full redraw. This can be achieved using the menu item Pairs>>Compound Alignment>>Use Rotation.