Extracting knowledge from protein structure geometry Peter Rogen, Patrice Koehl Department of Computer Science and Genome Center, UC, Davis Proteins 2013 Presented by Chao Wang
Background Model Energy Knowledge Estimation Generation vs. Evaluation Physics-based vs. Knowledge-based Knowledge Local vs. Global (Non-local): mean force Estimation RMSD, GDT_TS
Introduction
Methods Local Geometry: 7-mer fragments Nonlocal Geometry Solvent effects Weights training
Local
Ignoring Smooth Kernel
Non-local: A pairwise potential Ignoring regularization
Modeling Solvent Effects
Complete Potential
Discussion
Chao’s comments This potential can’t describe the first phase of folding. Hierarchical potential.