For uncommon NMR nuclei I often encounter a situation
where I have only one shift for a specific HOSE code.
It would be nice to have a default there. SK told me
that in case of C or H NMR, a CDK lookup table is used.
An alternative for other nuclei would be to calculate
the standard deviation (SD) for each HOSE code for each
sphere level. Then, for each sphere level a mean SD is
calculated, which will be large for 1 sphere and much
smaller for 6 sphere. This SD would be a nice default
for the SD for that HOSE code, as it at least will give
a likely region.
Currently, instead, the website just saying the given
shift is wrong (see screenshot) as it does not exactly
match the one known shift for that position.
These mean SDs would be calculated once every month or
so, as they are not overly dependent on the number of
new structures.
And, they also serve some educational use, as the
clearly show the effect of using many sphere's with
HOSE codes, and useful in generally accessing quality
of predictions.
