October 27, Mon 2014
1:00 pm, MRB 200
Dr. Jooyoung Lee
Center for In Silico Protein Science, Korea Institute for Advanced Study
Protein Structure Modeling by Global Optimization using Ambiguous NOE restraints
In the recent CASP (Critical Assessment of Protein Structure Prediction) 11 experiment, a new challenge, the contact assisted category (called Ts target) using simulated sparse NMR contacts was tested. Contacts based on simulations (carried out by the Gaetano Montelione's group) reflect the situation in the initial stage of the NMR experiment. For a fairly large protein (> 160 residues) for NMR, data is typically collected from deuterated samples, which usually results in much spectral overlaps that are difficult to assign properly. The number of simplified spectra straightforward to assign is rather small. Therefore, in the early stage of the protein structure determination by NMR, one is faced with the combinatorial optimization problem to properly assign ambiguous NOE peaks to their corresponding hydrogen atom pairs, which is followed by subsequent structure optimization satisfying all the distance constraint arranged by a given set of the peak assignment. With sparse constraints and the ambiguous distance information, standard NMR structure calculation programs fail to generate accurate protein 3D models, and the protein structure determination in such a situation remains as a challenge. In CASP11, 19 Ts targets with the chain length in the range 108 to 462 residues are tested. We solved the two-level optimization to generate 3D protein models, for which we have applied the global optimization method of conformational space annealing. In most of the cases, rather accurate 3D models were generated typically within 1.0 – 3.0 Å RMSD from the native structure. Proper application of the proposed method can greatly reduce the time and the cost of protein 3D structure determination using NMR data.