Why the correlation coefficient of the electron density with the model is never 1?
The links below show a plot of the results of phenix.real_space_correlation* for a nice lisozyme crystal and also the coot view. https://drive.google.com/open?id=0B9tqXyDsq9YIdkRhcVFXU2dEdnhINXR6Y3hza2Y5NG... https://drive.google.com/open?id=0B9tqXyDsq9YIREd1OFZhNGZtRnN4aldtVU1mdk84NE... As you can see the electron density fits very well the model. Why not a single residue gets a perfect correlation coefficient? (i.e. 1) Just curious. Thanks * phenix.real_space_correlation pdbfile mtzfile --detail="residue" > outpu
Hi Murpholino, reasons are similar to why crystallographic R factor never goes to zero. Model-map CC is calculated between the map (the data, in case of cryo EM, or sort of data in case of Crystallography) and a model. As you know data always has errors and model is just an approximation to reality. So you have no chance to match them perfectly, unless overfit. Something along these lines.. Pavel On 6/14/19 12:42, Murpholino Peligro wrote:
The links below show a plot of the results of phenix.real_space_correlation* for a nice lisozyme crystal and also the coot view. https://drive.google.com/open?id=0B9tqXyDsq9YIdkRhcVFXU2dEdnhINXR6Y3hza2Y5NG... https://drive.google.com/open?id=0B9tqXyDsq9YIREd1OFZhNGZtRnN4aldtVU1mdk84NE...
As you can see the electron density fits very well the model. Why not a single residue gets a perfect correlation coefficient? (i.e. 1)
Just curious.
Thanks
* phenix.real_space_correlation pdbfile mtzfile --detail="residue" > outpu
_______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]
Thank you for the clarification. El vie., 14 de jun. de 2019 a la(s) 14:50, Pavel Afonine ([email protected]) escribió:
Hi Murpholino,
reasons are similar to why crystallographic R factor never goes to zero.
Model-map CC is calculated between the map (the data, in case of cryo EM, or sort of data in case of Crystallography) and a model. As you know data always has errors and model is just an approximation to reality. So you have no chance to match them perfectly, unless overfit. Something along these lines..
Pavel
On 6/14/19 12:42, Murpholino Peligro wrote:
The links below show a plot of the results of phenix.real_space_correlation* for a nice lisozyme crystal and also the coot view.
https://drive.google.com/open?id=0B9tqXyDsq9YIdkRhcVFXU2dEdnhINXR6Y3hza2Y5NG...
https://drive.google.com/open?id=0B9tqXyDsq9YIREd1OFZhNGZtRnN4aldtVU1mdk84NE...
As you can see the electron density fits very well the model. Why not a single residue gets a perfect correlation coefficient? (i.e. 1)
Just curious.
Thanks
* phenix.real_space_correlation pdbfile mtzfile --detail="residue" > outpu
_______________________________________________ phenixbb mailing [email protected]http://phenix-online.org/mailman/listinfo/phenixbb Unsubscribe: [email protected]
participants (2)
-
Murpholino Peligro -
Pavel Afonine