#83 List of most wanted structures

[Egon Willighagen]

When doing a C-NMR predictions, the GUI now mentions
the number of spheres used in the prediction. It does
not state, however, why 'only' 1 sphere is taken. Some
indication why this is done would be welcomed. I guess
the reason is that for a 2-sphere (or higher sphere)
HOSE code, there are no associated peaks in spectra in
the database. BTW, this would be nice: a top-X of most
wanted X-sphere peak positions for a HOSE code based on
the predictions done on the website. This would require:
1. a log/hit list with requests per HOSE code (/per #
spheres, though that is implicitely in theHOSE code)
2. a web page that would list the most 'requested' HOSE
codes for which no peaks are found in the database
For the 1 sphere HOSE codes this list will be small
presumably, but for the 2-sphere this list is longer.
Such a list would help me prioritize which molecules to
add to the database. At this moment I am only at an
average rate of 1 spectra/day. I would insert those
spectra first which have HOSE codes which are not yet
in the database.

[Stefan Kuhn]

Logged In: YES
user_id=452972

I will put something about the working of the prediction
into the help. Your assumption is quite right.

[Christoph Steinbeck]

Logged In: YES
user_id=54358

We need to have an extensive database of real structures for
that. Possibly, this should be something like the dictionary
of natural products or so. The problem is that the set of
possible HOSE codes is likely to be much larger that the set
of acutally existing ones. The latter can only be accessed
via a database of real structures. We'll look for one...

[Stefan Kuhn]

Logged In: YES
user_id=452972

Note that there is a wishlist now in nmrshiftdb

[Egon Willighagen]

Logged In: YES
user_id=25678

I still would appreciate a list of 1/2/3 sphere HOSE codes
for protons with no associated experimental data yet. The
website specs.net has many H-NMR spectra, and doing a
substructure search on a specific HOSE code is a piece of code.