The 100k PDB biostructures, a unique source of Bioisosteric Replacement by Pocket Mining: further statistical analysis and validation of the FcBioisostere software solution
Bioisosteres in medicinal chemistry refers to structural changes on molecule that are not affecting existing biological activities. It’s a powerful tool to optimize the pharmaco-dynamic profile. Source of bioisosteric replacement are provided by (1) manual inspection of structure-activity literature (2) collection over such literature, (3) automatic analysis of Structure-activity data to detect chemical substitution that are maintaining the activity profile, (4) data-mining of 3D molecular data such as the CSD or the PDB databases.
Here we are exploring bioisosteric rules extracted by crossmining in 3D/2D the PDB and small Pubchem molecules to detect local pair of similar fragment-protein 3D interactions, as implemented in FC-Bioisostere software [Moriaud2011]. First, new statistics are measured on the overall distribution of those pairs of superposed PDB-based chemical moieties. 2D duplicates pairs of bioisoscteric replacement are detected and sorted according to cases where at least one pair is having the same functional annotation in the protein binding cavity, in order to score the chemical mutation suggested by other pairs having the same 2D fragments. Second structure-activitiy data related to bioisosterism mutation are qualitatively compared with FC-bioisostere pairs
This work on FC-Bioisostere explores one step further the overall chemo-proteomic challenge as initiated in our C2P Chemo-Proteomic Platform to better understand and predict interactions between, on one hand, ligands and all related fragments and, on the other hand, binding sites and all related subpockets.