A joint project of the Graduate School, Peabody College, and the Jean & Alexander Heard Library

Title page for ETD etd-11092011-164742

Type of Document Dissertation
Author Kaufmann, Kristian Wallace
Author's Email Address kristianwkaufmann@gmail.com
URN etd-11092011-164742
Title Computational prediction of protein small molecule interfaces using ROSETTA
Degree PhD
Department Chemistry
Advisory Committee
Advisor Name Title
Dr. Jens Meiler Committee Chair
Dr. Brian O. Bachmann Committee Member
Dr. Michael P. Stone Committee Member
Dr. Randy D. Blakely Committee Member
  • homology modeling
  • virtual screening
  • drug design
Date of Defense 2009-06-30
Availability unrestricted
Protein small molecule docking has focused on the modeling of small molecule flexibility and scoring of small molecules binding to fixed protein structures due to the inherent complexity of incorporating protein degrees of freedom. Recent developments in modeling of protein folding have opened the possibility of including protein degrees of freedom in small molecule protein interface modeling. ROSETTA, a protein modeling suite, has performed at the forefront of protein modeling in recent years. Its combination of knowledge based discrete sampling and knowledge based energy functions have pushed protein modeling to sub-angstrom accuracy.

In the dissertation existing ROSETTA sampling protocols and energy functions are discussed along with previous applications of Rosetta to a variety of protein modeling tasks including de Novo protein folding, comparative modeling, protein docking, and ligand docking with rigid small molecules. Expansion of ROSETTALIGAND to allowing simultaneous sampling of protein binding site flexibility and small molecule flexibility using a parallel knowledge based approach to both sampling and scoring is detailed. In a benchmark of small molecule docking the new method recovered a native-like binding mode in 9 of 10 cases when docked back into the parent crystal structure, while in 7 of 11 cases the protocol recovered a native-like binding mode when docked to a structure of the same protein crystallized with a different small molecule. The value of specializing energy scoring functions to specific ligand families is examined in the context of PDZ. Specialized energy functions are shown to improve prediction of binding energies upon mutation within PDZ domains and to predict specificity of peptides binding to the domains. We dock ligands to comparative models built by Rosetta and models from the 8th Critical Assessment of Structure Prediction. We find that 60% of cases produce a native-like model among the top ranked models indicating that comparative models can be used in predictions. Finally, an advanced case study in modeling a small molecule protein interface is described using serotonin bound to the serotonin transporter.

  Filename       Size       Approximate Download Time (Hours:Minutes:Seconds) 
 28.8 Modem   56K Modem   ISDN (64 Kb)   ISDN (128 Kb)   Higher-speed Access 
  Kaufmann_Dissertation_Chemistry_2011.pdf 4.75 Mb 00:21:59 00:11:18 00:09:53 00:04:56 00:00:25

Browse All Available ETDs by ( Author | Department )

If you have more questions or technical problems, please Contact LITS.