N Sukumar - Exploration of the Topology of Chemical Spaces with Network Measures

Version 1

      Publication Details (including relevant citation   information):

     

      J. Phys. Chem. A,115(45), 12905-12918 (2011).

      Abstract:

      Discontinuous changes in molecular structure (resulting from   continuous transformations of molecular coordinates) lead to   changes in chemical properties and biological activities that   chemists attempt to describe through structureactivity or   structureproperty relationships (QSAR/QSPR). Such relationships   are commonly envisioned in a continuous high-dimensional space of   numerical descriptors, referred to as chemistry space. The choice   of descriptors defining coordinates within chemistry space and   the choice of similarity metrics thus influence the partitioning   of this space into regions corresponding to local structural   similarity. These are the regions (known as domains of   applicability) most likely to be successfully modeled by a   structureactivity relationship. In this work the network topology   and scaling relationships of chemistry spaces are first   investigated independent of a specific biological activity.   Chemistry spaces studied include the ZINC data set, a qHTS   PubChem bioassay, as well as the space of protein binding sites   from the PDB. The characteristics of these networks are compared   and contrasted with those of the bioassay SALI subnetwork, which   maps discontinuities or cliffs in the structureactivity   landscape. Mapping the locations of activity cliffs and comparing   the global characteristics of SALI subnetworks with those of the   underlying chemistry space networks generated using different   representations, can guide the choice of a better representation.   A higher local density of SALI edges with a particular   representation indicates a more challenging structureactivity   relationship using that fingerprint in that region of chemistry   space.

      Address (URL):