Density Void in G(r)

[Emily S]

Hi Bachir,

I am simulating the PDF spectrum for a single cobalt molecule in a water solvent box (tot. ~2400 atoms).  The refined structure that I'm getting looks very reasonable, however, my G(r) has an odd profile (see attached).  Is this because of a density void?  As discussed in a separate thread, does the shape function need to be set accordingly, or is there something else I need to check here? 

Thank you,
Emily

[Bachir Aoun]

Hi Emily,

two things you might be having here. 

• you might be using the wrong pdf, fullrmc uses the Billinge one where i think your experimental data use the Keen formula. I will create both in the next fullrmc distribution
• In addition, i see a sudden drop in the intensity, this is because your simulation box half distance is smaller than the length of the experimental data.

you can always convert your experimental data to the one the fullrmc expects. read here to understand what fullrmc uses

http://bachiraoun.github.io/fullrmc/fullrmc.Constraints.html#module-fullrmc.Constraints.PairDistributionConstraints

thanks 

[Emily S]

HI Bachir,

Thanks for your reply.  I'm using PDFGetX2 to process my experimental data -- I believe this is the Billinge method unless I'm mistaken.  I'm not familiar with the Keen method and will read up on the difference between both.  

Right, there is a drop in intensity due to my constraints which may be the root of the problem.  I had initially set an outer coordination sphere boundary because we're not sure if the oscillations in my experimental data >12 A  are real, so I was concerned this may affect or distort my structure in a chemically unmeaningful way.  I'll try releasing the constraint and that should solve the issue.  

Thanks,
Emily

[Bachir Aoun]

Hi Emily,

one more thing, if you look at the small distances range, you have unphysical peaks in your model. I am sure you have input structure issues or boundary conditions issues ...

if you want, share with me your input file and i will help you set it up. or we can have a call at some point to explain to you what you should do.

regards 

[Emily S]

Yes, I've noticed that as well.  I was hoping that was a product of the solvent background, but I think my input structure does need to be modified.  The solvated structure was created in CHARMM as my collaborator is running MD simulations in parallel to my fullRMC fitting.  I then ran the solvated structure through pdbparser to prepare it for fullRMC.  I'll send you both files at your  ful...@gmail.com account.  Thanks for your kind offer to help.