A molecular pair table is built by analyzing a group of molecules to identify all pairs of molecules that are otherwise identical except for a change at a single position.  The idea behind this, is that when matched with data, a single change in the molecule can be attributed to a change in activity data. The goal is to remove all possible confounding structural variables to unequivocally connect a single structural change to a perturbation in the observable data. This of course ignores all sources of possible experimental error, hopefully small, that would add noise to any SAR analyses.

The question immediately arises as to what constitutes a relevant or actionable structural change. Certainly, in every medicinal chemistry series, one could build an arbitrary number of meaningless molecular pairs where the core is a single carbon atom or small number of atoms, and the rest of the molecule would constitute the changing fragment. An additional consideration is that the pair of molecules may fragment along lines that are incongruent with the synthetic chemistry of the molecular series. This leads to the question of what constitutes a relevant molecular pair for our purposes of analysis of structure vs activity. 

In SARvision|SM, the molecular pair relevance problem is solved by building molecule pairs that are anchored to a scaffold of interest selected by the user. The user can build as many pairs for as many scaffolds as desired simply by adding a molecule pair table (main menu->Insert->Substituent analysis) and selecting the relevant scaffolds one at a time (double clicking). The pair finding algorithm begins with a subset of molecules that belong to this scaffold and identifies molecular pairs that include this scaffold core.  The two pairs are shown side by side with structural differences denoted by different colors. Below each structure is observable data measured for each molecule. The third column focuses on the change in structure (top) and data(bottom) for this molecular pair. Note that the change in data can be depicted as difference or a ratio (table control: right bottom: drop down).

Often too many molecular pairs are derived for closely related sets of molecules and it is desirable to subset further by pairs that change at only specific positions. In the control panel (right) is a check box for each R-Group position (the R-Groups are the same as described in the R-Group table). Unchecking an R-Group temporarily removes it from the display allowing browsing of the data one scaffold R-position at a time. Note that molecular pairs are now filtered by a scaffold core and by position to make analyzing molecular pair changes easy to visualize.

Sometimes a series may have two closely related cores and the user may want to identify pairs of molecules that have the same R-Groups but change at the core. Placing these cores into a folder in the scaffold tree (scaffold pane:right click->Add folder, folder:right click->Add scaffold) and selecting the folder (folder: double click) creates a new molecular pair table. This is built listing only pairs where the cores listed in this folders are swapped. The table is otherwise identical to those created with a single scaffold. This can be useful for core hopping exercises.