- About SSRL
- SSRL News
- User Resources
- Beam Lines
- Science at SSRL
- Staff Resources
- Contact Us
Tuesday, November 13, 2012 - 11:15am
SSRL, Bldg. 137-322
Speakers: Julie Bernauer, Debanu Das & Dimitar Pachov
11:15-11:45 Julie Bernauer (INRIA AMIB Bioinfo) Multi-scale modeling for RNA structures: a challenge
11:45-12:00 Debanu Das (SSRL JSCG)
Progress on HT-SB of Protein/Nucleic Acid complexes at JCSG
12:00-12:15 Dimitar Pachov (SSRL and Stanford Chemistry Dept) Revealing conformational states of proteins by a novel model-building method for X-ray data
RNA molecules are involved in most biological processes and are therefore therapeutic targets of interest. The prediction of their 3D structures and interactions at different levels of representation is key in understanding their function. In silico studies and simulations mainly use physics-based potentials and techniques that could give structural insights in atomic detail but with very limited accuracy, particularly in predicting native like molecular structures of RNA molecules and their complexes. Recent efforts in RNA structure prediction techniques have shown that parameterized energy functions and knowledge-based techniques largely improve the accuracy of structure prediction.
In our project, we showed that distance-based RNA KB potentials could be build both in an atomic and coarse-grained setting using mixture models. The obtained potential performed better than the existing tools. We also showed that different knowledge-based strategies (mixture models, machine learning or hybrid potential functions) improve the predictions. Due to the ruggedness of KB potentials, it is unclear how these will perform in minimizing structure and sampling different conformations, especially in a hierarchical setting. The junction problem should thus be addressed to study large RNA molecules.
We hope our recent models using biogeometry, statistics and machine learning will allow to improve RNA structure modeling, possibly dynamics simulations, junction characterization and protein-RNA interactions.
The PSI-Biology program is a networked High-Throughput Structural Biology platform for determining the 3D structures and analyzing the structure-function relationships of proteins and their complexes that are of substantial biomedical and biological importance. The JCSG (at SSRL Structural Genomics Division) is one of the HT Structure Determination Centers that is in partnership on several projects aimed at elucidating the structures of proteins and protein-nucleic acid complexes that are involved in stem cell biology and regulation of T-cell activation. A brief summary of the current progress will be presented.
Biomolecules operate by switching between functional dynamic states. Such flexibility carries important functional implications, but it also leads to conformational heterogeneity which reduces quality of X-ray experimental data and imposes difficulties in model fitting and building of large molecules. We present development of an innovative, robotics-inspired, automated method to build high quality structural models of large complexes into crystallographic data while preserving the characterization of its conformational substates.