| Home | Documentation | I/O Formats | Gallery | Contact |
TablesTable I: GeNMR knowledge-based score (SumGA) can distinguish high-resolution X-RAY structures from lower resolution or lower quality NMR models of the same protein.
Table II: GeNMR chemical shift scores (SumCS) of high-resolution X-RAY structures and lower resolution/lower quality NMR models of the same protein.
Table III: Changes in RMSD of homology models after refinement with distance restraints and GeNMR scoring function.
Table IV: Changes in GeNMR scoring function after refinement homology models with distance restraints and GeNMR scoring function.
Figure 1. Restoration of secondary structure in homology models by a refinement with GeNMR scoring function without NOEs1. Ubiquitin modeled from 2HJ8 after the 450 iteration of refinement with GeNMR scoring function. A. Reference structure (1UBQ). B. Homology model. C. Refined model. 
2. Ubiquitin modeled from 1IYF after the 683 iteration of refinement with GeNMR scoring function. A. Reference structure (1UBQ). B. Homology model. C. Refined model. 
3. Ubiquitin modeled from 1OQY after the 490 iteration of refinement with GeNMR scoring function. A. Reference structure (1UBQ). B. Homology model. C. Refined model. 
Table V. Performance of GENMR in generating 3D structures for 50 randomly selected BMRB files. The given RMSDs are those of the final structure relative to the known (X-ray or NMR) structure.
Table IV. Backbone RMSD between GENMR models and corresponding experimental structures for recent entries in BMRB database (as of the first week of April 2008).
Table V. Backbone RMSDs between models generated by GENMR sub-program for ab initio folding and corresponding experimental structures.
Table VI: Ubiquitin models built with GENMR using (a) homology modelling, (b) threading, and (c) ab initio folding. RMSDs and sequence identity values were calculated with respect to the native structure (1UBQ). Tables (a) and (b) show the ubiquitin models built as ubiquitin homologs were progressively removed from the database, so that the homolog used in a given row is the best one found by GENMR when the more closely matching homologs above it were removed. (c) Results of ab initio folding of ubiquitin by Rosetta. 50 models were produced in each of 10 runs. Backbone RMSDs of the model with the best GENMR score are shown, with those less than 2.8 Å labeled as successes. Homologous proteins were excluded from Rosetta's fragment generation step. (a) Homology Modelling
(b) Threading
(c) Ab initio
Table VII. Comparison of model global accuracy achieved by GENMR and similar programs.
(b) Comparison of backbone RMSDs of models generated by CS-ROSETTA and GENMR for proteins that did not meet CS-ROSETTA convergence criteria.
(c) Comparison of backbone RMSDs of models generated by CS-ROSETTA and GENMR for structural genomics proteins.
Figure 2. Dependence of backbone RMSD of GENMR models on template sequence identity for different methods of structure generationA scatter plot showing the performance of GENMR three approaches to structure generation (subfragment assembly + homology modeling, chemical shift threading and ab initio structure generation) relative to the level of sequence identity of the matching templates or subfragments. Data from Tables III-VI in the GENMR web documentation pages was used to assemble this graph.
Energy FunctionSummaryGENMR uses an energy function consisting of several statistics-based pseudo-potentials:Score WeightingEach of the initial terms in the energy function are transformed through a series of steps: (1) They are multiplied so that they behave as energy-like terms, i.e. more negative scores indicate a better structure. (2) They are scaled so that they all have close to the same order of magnitude (about ± 1 to 100). (3) All scores except for the chemical shifts scores and the radius of gyration score are normalized by dividing by the number of residues in the peptide, and then multiplying by 100 to re-scale them. (4) All scores are multiplied by weights that were determined based on training of the energy function on CASP7 decoys and decoys generated by Rosetta.The initial score calculations, plus the results of steps (1) and (2) are: The weightings for step (4) for the above scores, other than the chemical shift scores, were trained on decoys from CASP7 using linear regression-based fitting. The weights for the chemical shift scores were trained separately on Rosetta decoys of structures with known chemical shifts; both ab-initio Rosetta decoys and Rosetta decoys generated from relaxing the native structures. The relative weighting between the chemical shift scores and the other scores was then determined based on linear regression. The weights in step (4) are multiplied by the values in step (3) to obtain the final scores. The step (4) weights are:
† Adjusted manually to a low value because training set it to 0. GENMR Results AssessmentA GENMR results summary for a structure prediction includes the mean chemical shift correlation values before and after energy minimization. The reliability of the final structure is guaged by looking at the final mean chemical shift correlation:0.75 - 1.00 = High 0.65 - 0.75 = Good 0.55 - 0.65 = Moderate 0.00 - 0.55 = PoorA GENMR results summary also prints information on torsion angle violations. The expected number of torsion angles in the core, allowed, generous and disallowed regions is estimated based on the following percentages: #res in phi/psi core 90% #res in phi/psi allowed 7% #res in phi/psi generous 1% #res in phi/psi disallowed 0% #res in omega allowed 99% #res in omega disallowed 1%Note: The first and last residues are not use for phi/psi, and the last residue is not used for omega. Torsion Angle Mapping for ThriftyGENMR performs a homolog search based torsion similarity against a non-redundant database of all cytoplasmic proteins in the PDB. This search uses an updated version of Thrifty that assigns each residue a letter based on its torsion angle class. The torsion angle mappings for each of the 9 letters in this torsion angle alphabet are shown below.
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
