Selectivity mining – multiple activities in Activity Miner

In drug discovery, selectivity is often as important as potency. Many potent drugs cannot be used due to their lack of selectivity.

Activity Miner was released in 2013 to analyze and visualize 3D activity cliffs. The next release includes the analysis and visualization of 3D selectivity cliffs and was presented and demonstrated by Martin Slater, Director of Consulting at Cresset.

For each pair of molecules the activity difference can be defined as the difference between activity divided by the distance. The distance is defined as being 1 minus the similarity.

2D similarity can cause problems as it can fail to identify bioisosteres and fail to deal with chirality.

3D molecular similarity is defined for conformers, not for molecules. Each compound can generate a large number of conformers, therefore the context for the 3D similarity is important – if you know how it binds with the protein then the comparison will be far more meaningful.

We have effective tools for aligning and scoring molecules. The challenge is how to visualize the results. Cresset has developed a number of methods.

Top Pairs is a tabular method of looking at the highest disparity pairs.

The Disparity Matrix is a sortable table of all of the results. There is high colouration of the large disparities, identifying a drop off or increase in potency.

The Activity View is an innovative way of comparing the similarity between a chosen focus molecule and its closest neighbours. The fields electrostatic environment can also be added and this can be adjusted so that the fields difference can be seen. This makes it possible to see the electrostatic changes that result from the changes in structure. These difference plots show you the regions where each molecule has stronger electrostatics.

Analyzing and visualizing selectivity cliffs

Selectivity cliffs make it possible to look at the structural changes that cause large changes in selectivity. Again, the challenge is not so much the calculations as the visualization.

The Activity View and Disparity Matrix have been changed to show multiple activities: the coloured display is split up into more and more areas to show the different colour coded activities.

This was followed by a demonstration of the software.