Hello Philippe-- > I am trying to run ensemble simulation with noe and SAXS data. I > used an adapted input from the xplor distribution (ensemble.py from > the dna_ensemble tutorial) and things went quite well now. However i > did not find the exact way that cross validation of the SAXS data > works. I only see in the input the pot terms that are cross > validated and find corresponding indication in the header of the pdb > files produced, but i did not succeed in locating the (presumably) > default parameters used for the cross validation (like the fraction > of data used for cross validation and so on). Can anyone give a way > to find the parameters associated with CrossTerms and cross > validation in the xplor-nih documentation ? Or a > detailed description of it in a paper ?
The xrayCross term in the dna_ensemble example does not constitute an independent cross-validation- it's simply a check as to whether the curve calculated with 30 q points gives the same behavior as that calculated with 61. The dna_ensemble and other examples included in eginput don't include SAXS terms for true cross-validation. Cross-validation can be achieved by leaving the SAXS term out of the potList. Another approach sometimes taken is to use only small-angle SAXS information in the fit, and cross-validate against the curve at larger angles. This could be done within an Xplor-NIH structure calculation script, but can also be done after the fact using the calcSAXS helper and the calculated structures. Please contact me if you'd like assistance with this. best regards-- Charles _______________________________________________ Xplor-nih mailing list [email protected] https://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
