Best practices for solvent density and hydration shell

[sammahdi]

I am attempting to calculate the theoretical SAXS profile using Crysol for a single PDB. I generally use 0.388 for the solvent density due to TI3P model, but I know a range of values can be used and accepted. The same can be said of the hydration shell. However, say in the absence of experimental data, how does one know which is the best values to be used for the solvent density and hydration shell? A one size fits all can't be used, so if I decide to look at other structures, how do I know what a good value should be? Is there a standard values for both for globular proteins (are the default values picked in crysol those standard values)?

Additionally, a question on the .int file. I notice there is theoretical intensity in solution, solvent scattering, an border layer scattering. What is the difference between all of these? Is the solvent scattering just the theoretical scattering from the solvent (which I presume is already subtracted in the theoretical intensity, so why is it given?)? I just want to know which one is the theoretical scattering profile using the given solvent density and hydration shell.

Finally, where is the errors for the theoretical profile? I can't seem to find any output that gives the error from this model? And I also noticed in .abs the intensities are normalized to concentration, but I haven't given a concentration, so how is this being generated? Is there a default built in concentration, if so what is it? I can't seem to find info on this in the documentation.

Thank you ahead of time!

[Guðmundsdóttir]

Additionally, a question on the .int file. I notice there is theoretical intensity in solution, solvent scattering, an border layer scattering. What is the difference between all of these? Is the solvent scattering just the theoretical scattering from the solvent (which I presume is already subtracted in the theoretical intensity, so why is it given?)? I just want to know which one is the theoretical scattering profile using the given solvent density and hydration shell.

I guess the "solvent scattering" (4th column) is the scattering of the displaced solvent volume (with the given solvent density dns). The total theoretical scattering profile is in the second column, should be the same as the second column in the .abs file but in some internal units of CRYSOL.

Finally, where is the errors for the theoretical profile? I can't seem to find any output that gives the error from this model? And I also noticed in .abs the intensities are normalized to concentration, but I haven't given a concentration, so how is this being generated? Is there a default built in concentration, if so what is it? I can't seem to find info on this in the documentation.

Eh, why would you expect experimental errors on a theoretical profile?
I guess the concentration of the .abs file is equal to one mg/ml.

Sorry, no idea how to estimate the solvent density and the contrast of hydration shell. For proteins in an aqueous buffer I just stick to the default values.

[sammahdi]

I ended up just running a grid search of solvent density and hydration shell to see which fit my data best. Turned out to be 0.3475 for the density and 0.055 for the shell (default values gave me a worst fit).

[Guðmundsdóttir]

Grid search is a cool idea, thanks for sharing.
I think that CRYSOL will actually find the best fitting hydration shell contrast if you don't provide the --dro value.

[sammahdi]

While that is true, I've found solvent density and hydration shell vary from run to run. Take for example multiple SAXS runs at difference concentrations, but only due to a single species. Theoretically, that single species should have the exact same dro/dns values for each run (concentration should have little influence on dns/dro), yet I've found I can get significantly different dro/dns values between individual fits. So to resolve this issue, I instead do a global fit, where I fit my data at all concentrations and minimize the dro/dns values.