ORGN 630 |
| Many drugs are recognized by seemingly unrelated targets, at least as measured by bioinformatic metrics. Chemically, however, the drugs themselves are typically related. It should thus be possible to relate protein targets based on the similarities among the ligands that bind to them. To investigate this, we used sets of ligands annotated for several hundred targets, comparing the topological similarity of every ligand across every set. A statistically significant similarity score for each pair of ligand sets can be calculated once a model of random similarity is developed. A minimum spanning tree can be found that maps all the sets together using their most significant links. Although no biological information is used in calculating these maps, biologically sensible clusters nevertheless appear as an emergent property. Relating ligand sets, and the mapping it enables, may reveal pharmacological effects and mechanisms for new chemical entities. |
|
Chemical Information and Organic Chemistry: The Road Ahead
8:25 AM-12:00 PM, Wednesday, 13 September 2006 Moscone Center -- Room 131, Oral
Division of Organic Chemistry |