Ca occupancy in multiple soaked structures
Hi Everyone, I have a complex case of data analysis I would love some opinions on. I have a group of structures of a protein that binds 6 calciums, and we want to know the order/concentration at which they fill. The crystals were soaked with calcium from 0-10 mM and all are 1.6-2.1 A in resolution. I determine the order of the calcium binding, and now I am working on the percent occupancy at each concentration. To do this, I have removed the calciums, and then performed simulated annealing for each structure from 2.1-30A resolution to 'normalize' them. I then look at the omit map and scroll out to find the peak sigma intensity of where the calciums should be. The curves fit very well for each calcium, and the numbers correspond well with other solution studies we have. My thought is that I need to normalize this in another way, so I thought choosing a water molecule that is present in all of the structures and removing it, then performing simulated annealing would be a good idea to normalize the peak intensity for each calcium soak concentration, since this water molecule shouldn't change occupancy. Is this the proper way to do this, or can you guys suggest a better way to go about this. I also looked at B factor and e/A3, but that isn't as consistent as peak sigma. Thanks in advance. Dan Daniel J. Slade, Ph.D. Research Associate Thompson Lab The Scripps Research Institute 130 Scripps Way Jupiter, FL 33458 (561) 228-3213
I would say e/A3 is physically more meaningful, even if it doesn't fit the theory that well...peak sigma depends more on the flatness of the rest of the map. On 22 May 2013, at 19:10, Daniel Slade wrote:
Hi Everyone,
I have a complex case of data analysis I would love some opinions on. I have a group of structures of a protein that binds 6 calciums, and we want to know the order/concentration at which they fill. The crystals were soaked with calcium from 0-10 mM and all are 1.6-2.1 A in resolution. I determine the order of the calcium binding, and now I am working on the percent occupancy at each concentration. To do this, I have removed the calciums, and then performed simulated annealing for each structure from 2.1-30A resolution to 'normalize' them. I then look at the omit map and scroll out to find the peak sigma intensity of where the calciums should be.
The curves fit very well for each calcium, and the numbers correspond well with other solution studies we have. My thought is that I need to normalize this in another way, so I thought choosing a water molecule that is present in all of the structures and removing it, then performing simulated annealing would be a good idea to normalize the peak intensity for each calcium soak concentration, since this water molecule shouldn't change occupancy. Is this the proper way to do this, or can you guys suggest a better way to go about this. I also looked at B factor and e/A3, but that isn't as consistent as peak sigma.
Thanks in advance.
Dan
Daniel J. Slade, Ph.D. Research Associate Thompson Lab The Scripps Research Institute 130 Scripps Way Jupiter, FL 33458 (561) 228-3213 _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
I did something similar back in 2003 [see: JBC 278(14):12271-12277 {2003}) The occupancy and the B-factors are going to affect each other in the refinement. Refining both the occupancy and the B-factor parameters at the same time is what you want to avoid. The refinement programs want to believe you know what you are doing (that you know the occupancy in this case). Thus, you need to refine in cycles. First refine the occupancy, then refine the B-factor, then the occupancy, and then the B-Factor of each individual Ca2+ atom. Don't try to rush this by doing multiple atoms or automating the runs, you spent how many hours collecting all this data, go slow and really study each Calcium site. Once you think you have the right B-Factor and Occupancy value, try nudging one of the values up or down to see if you are in a local minima or if you really have the answer you are looking for. Steven Herron [email protected] On 5/22/2013 11:10 AM, Daniel Slade wrote:
Hi Everyone,
I have a complex case of data analysis I would love some opinions on. I have a group of structures of a protein that binds 6 calciums, and we want to know the order/concentration at which they fill. The crystals were soaked with calcium from 0-10 mM and all are 1.6-2.1 A in resolution. I determine the order of the calcium binding, and now I am working on the percent occupancy at each concentration. To do this, I have removed the calciums, and then performed simulated annealing for each structure from 2.1-30A resolution to 'normalize' them. I then look at the omit map and scroll out to find the peak sigma intensity of where the calciums should be.
The curves fit very well for each calcium, and the numbers correspond well with other solution studies we have. My thought is that I need to normalize this in another way, so I thought choosing a water molecule that is present in all of the structures and removing it, then performing simulated annealing would be a good idea to normalize the peak intensity for each calcium soak concentration, since this water molecule shouldn't change occupancy. Is this the proper way to do this, or can you guys suggest a better way to go about this. I also looked at B factor and e/A3, but that isn't as consistent as peak sigma.
Thanks in advance.
Dan
Daniel J. Slade, Ph.D. Research Associate Thompson Lab The Scripps Research Institute 130 Scripps Way Jupiter, FL 33458 (561) 228-3213 _______________________________________________ phenixbb mailing list [email protected] http://phenix-online.org/mailman/listinfo/phenixbb
Hi Dan, as few random thoughts: - I guess this only makes sense to do relatively: say pick one as 1 and see the others as fractions of it. Though Fourier maps need to be computed using the same sets of Miller indices for comparisons to be meaningful. - Remember you look at Fourier images of electron density, not electron density itself... That's the key. - All in all this means peak heights may mean more than simply site occupancy. - If you go by residual peak parameters (heights, for instance) then it doesn't matter how you refined occupancy and B of these Ca, and in which ways they are correlated. Pavel On 5/22/13 10:10 AM, Daniel Slade wrote:
Hi Everyone,
I have a complex case of data analysis I would love some opinions on. I have a group of structures of a protein that binds 6 calciums, and we want to know the order/concentration at which they fill. The crystals were soaked with calcium from 0-10 mM and all are 1.6-2.1 A in resolution. I determine the order of the calcium binding, and now I am working on the percent occupancy at each concentration. To do this, I have removed the calciums, and then performed simulated annealing for each structure from 2.1-30A resolution to 'normalize' them. I then look at the omit map and scroll out to find the peak sigma intensity of where the calciums should be.
The curves fit very well for each calcium, and the numbers correspond well with other solution studies we have. My thought is that I need to normalize this in another way, so I thought choosing a water molecule that is present in all of the structures and removing it, then performing simulated annealing would be a good idea to normalize the peak intensity for each calcium soak concentration, since this water molecule shouldn't change occupancy. Is this the proper way to do this, or can you guys suggest a better way to go about this. I also looked at B factor and e/A3, but that isn't as consistent as peak sigma.
Thanks in advance.
Dan
participants (4)
-
Daniel Slade -
Mark van Raaij -
Pavel Afonine -
Steven Herron