Spark databases

The currently released Spark fragment databases are derived from multiple sources:

• Commercially available screening compounds (eMolecules screening compounds)
• Literature reports (ChEMBL)
• Patent data (SureChEMBL)
• Commercial reagents (eMolecules building blocks)
• Theoretical ring systems (VEHICLe)
• Larger fragment databases are split based on the frequency of occurrence of the fragments.

Spark fragment databases based on small molecule crystal structure data from the Crystallography Open Database (COD) and Cambridge Structural Database (CSD) are also available.

Spark’s database generator allows you to create your own databases of potential bioisosteres using three different fragmentation approaches:

• 'Whole molecules' where the fragmentation process breaks each molecule into smaller, synthetically feasible pieces
• 'Pre-labelled fragments' loaded ‘as-is’, and not further fragmented, with the attachment points based on the chosen attachment point label
• 'Reagent importer' where molecules are processed according to the chosen transformation rule

Once formed, the fragments/reagents are conformationally searched, and the resulting conformations are stored in the database.