Sneak peek at Spark V10.6

In this blog post, I give a sneak peek at some of the new features in the upcoming release of Spark™, our bioisostere replacement tool.

Improved calculation methods

For the Spark V10.6 release we have significantly increased its performance on finding bioisosteres with the traditional Spark search methods (based on ligand electrostatic and shape similarity). Improvements include:

• Higher number of fragments passed to the final stages of scoring
• New ‘Very Accurate but Slow’ method using shape similarity scoring also in the initial stages of the search
• Searches starting from a protein-ligand complex using the protein as an excluded volume more efficiently.

Our validation experiments show that, in a typical R-group replacement experiment (Figure 1), these improvements can lead to the identification of up to 30% more results with promising BioIsostere Factor (BIF%).

Figure 1: R-group replacement experiment on PDB: 1OIT. Additional interesting results found using the improved calculation methods in Spark V10.6.

New docking method

A new search method, based on the Lead Finder™ docking algorithm, has been added to Spark V10.6. With the new ‘docking’ method, Spark will find results picking protein-ligand interactions directly from the protein structure. When you want your ligand to grow beyond the interactions made by the starter molecule you can use this method in ligand growing (Figure 2) and R-group replacement experiments. This method may also help you identify linkers which make specific interactions with the protein active site in ligand joining and macrocyclization experiments.

Figure 2: Ligand growing experiment on PDB: 1OIQ, using the new docking calculation method.

This new search method combines the user friendliness of Spark with the ability to generate high-quality binding poses of Lead Finder. During the search, each potential replacement fragment is merged to the starter molecule in a sensible initial conformation, which is then optimized by Lead Finder to generate the final pose for the result molecule. Critical protein-ligand interactions can be mapped by docking constraints to ensure that they are matched by the Spark search results.

Many other improvements

This release delivers new science together with significant improvements to the user interface, for example:

• Enhanced 2D pictures in all the result tables
• Improved advanced search options, to facilitate the setting of appropriate physico-chemical properties filters during the search
• Ability to distinguish between the different attachment points in the results from scaffold hopping experiments
• Pre-defined filters for common chemical groups and functionality such as Ring, Aromatic Ring, Chiral atom, H-bond donor/acceptor
• Import/Export of filters, customized calculation methods and radial plot properties
• All-new fragment databases, with a total of more than 7.6M fragments to search.

Stay tuned

Spark V10.6 is coming soon. A practical sneak peek will be given in the webinar ‘Escape prior art: Find accessible new chemical matter for your project’ which takes place on October 21^st – find out more and register to secure your place.