Read on Twitter
I& #39;m looking at the newly-released 7btf ( https://www.ebi.ac.uk/pdbe/entry/pdb/7btf)">https://www.ebi.ac.uk/pdbe/entr... - and something seems wrong in the vicinity of residues 910-920 of chain A (NSP12). The sequence here really doesn& #39;t appear to fit the density (look at the sidechains). Same complex as 6m71... will cross-check.
Hmm... now that I look at it again, the density looks similarly suspicious in 6m71:
... and in 6nur (the same complex from SARS - pretty clearly the starting model for the new ones, I think):
The problem here is that this helix is flanked on both sides by unresolved loops, so there& #39;s only about 10 residues worth of really good density to dock the sequence in to. It *looks* wrong, but what& #39;s the right answer? Will see if I can work it out.
Oh, wow - looks like they& #39;re all out by *nine* residues (Pro 918 is where Pro 927 should be; Leu 907 is where Trp 916 should be). Would *not* have been easy to pick this in the original 6nur, but 7btf is both higher resolution and has slightly more "tail".
OK, 7btf rebuilt end-to-end and uploaded to https://drive.google.com/open?id=1_DNCD4SsqZusopS_NkHFF2X49Ec67tNr.">https://drive.google.com/open... Will try to work back and correct the related models. Masked map-model correlation (from phenix.model_vs_map) increased from 0.8436 to 0.8464; MolProbity score down from 1.60 to 0.65.
Incidententally, this all makes 6nus (the earlier SARS model with the same register error) my new record-holder for stellar-looking validation summary over a serious problem. Again, no blame - that region of the density *was* difficult - but remember to check for yourself.
Being honest with myself, I don& #39;t think I would have been comfortable making this change in 6m71 without having seen it in 7btf first. In 7btf (left) the refitted Trp916 is *really* clear, but in in 6m71 it& #39;s... not so much. Looks like it has multiple conformers.
Nevertheless, the rest of the helix does clearly fit much better just as in 7btf, and the better density in the latter really does put it beyond question in my opinion.
6nur also refits beautifully. 6nus is no help, though (overall resolution much lower, and this region probably wouldn& #39;t be buildable at all on its own).
