Hi Charles, It appears to work well. The output file is a bit hard to read as each RDC value is its own group, so had to include a custom output script in the refine.py file. But, otherwise the structural calculation does appear to consider residue weights appropriately.
In most cases, I imagine the RDC uncertainties are relative gaussian, and can be considered a singular group. However, in cases with heterogenous line-broadening (such as mine), per-restraint groupings may be a useful feature to have available. Thank you for the help and for the paper. Bests, John On Fri, Dec 15, 2017 at 5:02 PM, Charles Schwieters <[email protected]> wrote: > > Hi John-- > > > Looking through the refine.py file, I see there are weightings for > different > > nuclei, which correspond to their associated gamma1*gamma2 values. > > If I understand you correctly, I can do something similar by simply > splitting my > > data based on SNR? For example: > > > > rdcs = PotList('rdc') > > for (medium, expt, file, scale) in [('Phage', 'NH_1', > 'RDC_High_SNR.tbl', 1), > > ('Phage', 'NH_2', 'RDC_Low_SNR.tbl', 0.1)]: > > rdc = create_RDCPot("%s_%s"%(medium,expt),file, media[medium]) > > rdc.setScale(scale) > > .... etc > > Correct. > > > > > If I'd like per residue weighting, could I then create a tbl file each > residue > > with scaling proportional to (1/uncertainty)? Something like: > > > > for Res, Unc in zipped(Residue_List, Uncertainty_List): > > rdc = create_RDCPot("%s_%s"%('Phage', 'NH_{0}'.format(Res)) Res + > '.tbl', > > media['Phage'] > > rdc.setScale((1/Unc)) > > Since the energy is proportional to rdc^2, you would usually use > 1/uncertainty^2. > > There is also the option of adding per-restraint weighting. It would > be quite easy for me, but time is short at the moment. Perhaps I could > get to this by the end of the calendar year. Even if we don't use it > by default, it could be a useful feature. > > > I'd consult the references you attached, but sadly my institution does > not > > access. > > > > That doesn't make me happy. I've placed an unproofed version here: > > https://nmr.cit.nih.gov/xplor-nih/methodsMolBiol.pdf > > best regards-- > Charles > -- *John B. Stiller* Ph.D. Candidate Kern Group Department of Biochemistry Brandeis University
_______________________________________________ Xplor-nih mailing list [email protected] https://dcb.cit.nih.gov/mailman/listinfo/xplor-nih
