Refinement Flashcards
What is the R factor?
This the difference between the Fobs and the Fcalc
(Fobs - Fcalc)/Fobs
What does the R factor do?
Indicates the fit of the model to the experimental data and identifies overfitting.
Do you want the R factor to be low?
Yes
What is an ideal R factor?
20%
What leads to overfitting?
More parameters than observations - this can lead to an misleading agreement between model and experiment
How many parameters does each atom have?
4 - xyz and B factor
What is the B factor?
This describes the atoms vibration about its mean position
What happens if an atom has a high B factor?
The electron density is smeared because the contribution of the atom to the scattering factor is reduced.
What can be included to increase number of observations?
Bond length, bond angle, dihedral angles, chiral centres, planar groups (e.g. F must be planar) and VDW contacts.
Does a correct structure need to fit the data or match prior knowledge?
Both
What is used to minimise stereochemistry differences between model and data?
A function that minimises the differences between Fobs and Fcalc
How does Maximum likelihood refinement work?
It weights hkls on their agreement with Fobs and Fcalc.
Describe the use of the Rfree factor?
Remove 10% of reflections at beginnig of model building (cross validation set). Use 90% (working set) to do model building. The working set is used to minimise Rwork when refining, Rfree is unaffected by overfitting of the data - so if Rwork decreases and Rfree does not then there is overfitting.
Does the model need to match the Ramachandron plot at the end of model building?
Yes
What does the R.M.S.D mean?
This is the deviation from perfect bond lengths and angles (want it to be low).
