Transcript Julia
Crystallographically determined 1. Holo: Complexed conformation 2. Apo: Uncomplexed conformation (Homology) Modeled Julia Salas Case Study, 3-14-06 CS379a Introduction • Docking often uses a single receptor structure – Some algorithms allow for reception flexibility but at high cost • For many targets, many structures are available • Most proteins don’t have published structures – 5,500 out of 730,000 with known sequence (2003) • 10 enzymes with known holo, apo, and homology structures • MDL Drug Data Report database molecules docked – 95,000 compounds…known ligands between 0.3-1% • DOCK3.5 used, ligands were allowed flexibility Enrichment factor was measured and compared – (# of known ligands found in ranked list) / (# expected to be found randomly) What is the influence of protein structure on docking success? How good does a receptor structure have to be for successful docking? Results In general, success of docking was best for holo followed by apo and lastly by modeled targets Limitations of each type of target structure: • Holo (Best in 7 systems) • Overspecialization • Ligands with very different binding geometries were not well docked • Apo (Best in 2 systems) • Structure may be very different from ligand-bound structure • Modeled (Best in 1 system) • Sidechains could be poorly placed Holo Active Site Docked in Holo Apo Active Site Docked in Apo EXAMPLE: Thrombin (a Serine Protease) •Apo was the best structure (24-fold over random): Slightly larger site •Holo had at best 19-fold over random: Misses some H-bonds, restricted structure Conclusions •Nearly all structures led to enrichment •Enrichment of at least 20-fold: 8 holo, 2 apo, 3 modeled •Holo structures are most likely to lead to useful enrichment Except: Overspecialized Holo structures •“Promiscuous” structures are the best targets •Flexibility allowances in binding pocket lead to more accurate results •Take into account the best of both the Holo and Apo structures