On 8 August 2016 at 21:04, Mahdi, Sam <sam.mahdi.846@xxxxxxxxxxx> wrote:
Hi, I had some general questions regarding model-free analysis that came up
as I was running it. About 2 months ago, I ran model-free on a 126 residue
protein. The pdb file I used for the protein had an extra 7 residue
modifier at one of the terminals, which made it 133 residues; thus, my data
did not match the pdf file. That run took a little over a week, with the
sphere model running about 2 models, along with every other model being
about 2. The data for this run I obtained was unusable (values as small as
5x10^-30 for S^2). I attributed this to the fact that my pdf files sequence
did not align with my data.
Do you mean that the residue numbers were out by 7? I have considered
this, but relax currently has no way of renumbering the residues in a
3D structure or renumbering the residues of the input data. So
unfortunately you'll either have to use a molecular view to change
your PDF file, or a script or spreadsheet to renumber your input data.
I adjusted my relax data (simply added 7 residues with no values), and ran
it again (all settings the same as before, everything automatic).
I don't think that your data will match up to the correct residues.
The residue numbers in the PDB file and the input relaxation data must
be exact matches.
This time
it has taken almost a month, with the sphere model on its 20th run. I saw
in a previous email that the max is 30 iterations. Is it normal to hit the
max? Does this mean my data is bad if it requires the max 30 iterations?
Also, why is it running up to 20 now? The 7 residues I added didn't have
data, so it doesn't run models for them. I assumed there would be a
difference in time since the data is now attributed to different spins, but
not this much of a difference.
If the 3D info and input data do not match, then the results for
non-isotropic (ellipsoidal and spheroidal diffusion) will make no
sense. In relax you should never need to add relaxation data for
these 7 residues, as relax will handle this situation automatically.
What does your input relaxation data look like?
Regards,
Edward
Re: Troubleshooting model-free analysis