There is some interest here at Illinois in acquiring a 1H and 13C
chemical shift, etc. NMR database. I think Stadtler sells one, but have
no experience with this type of software. Has anyone purchased,
demo-ed, used, etc a database of this kind? Suggestions as to vendors
would also be appreciated.
Many thanks to all of those who replied. The replies fell into two
categories - 13C and 1H spectra. Basically, there are databases for
13C, but most leave a lot to be desired for 1H. The replies are posted below,
separated into comments about specific programs.
SUMMARY:
CARBON-13
1) CSEARCH by Stadtler
a) CONTACT INFO: This letter is a response to your inquiries
about NMR chemical shift databases. Sadtler Division of Bio-Rad
Laboratories makes a carbon NMR database of 60,000 chemical shift
assignments available with its CSEARCH software product. The
product is available for Sun SPARC, Silcon Graphics and VAX VMS
hardware. I'd be pleased to send you more information. Rod Farlee,
our Development Manager, and I will be at 1994 ENC with the
CSEARCH system. Perhaps we'll see you there.
Kind regards,
Bruce Woods
CSEARCH Product Manager
Sadtler Research Laboratories
3316 Spring garden Street
Philadelphia, PA 19104
phone 215 382-7800
fax 215 662-0585
email bruce@sadtler.com
b) bio-rad sadtler division 215-382-7800 fax 710-670-1186 sells a
pc version called csearch with 7500 cpd for $2,000/yr or a workstation
version with 60,000 cpd for $90,000 (which is why i haven't gotten too
far yet). i think the pc version can go up to 30,000 cpd. there is no
internet access.
c) A couple years ago I sat in on one of the demos of the laptop
Sadtler data base software. It was pretty slick--you could tie in to their
data base by typing in a structure, name of a compound, or even a
fragment of a structure. It allowed you to literally draw it in. Then it
would display several compounds that fit that bill, and would allow you
to check out their spectra. The NMR data base also tied in with an IR
and UV data base. They were planning to add mass spec as well. It also
had the flexibility to allow you to draw up your own data base and
search it, as I recall. However, it was very expensive---$25-30k.
d) CSEARCH is the database that Sadtler are presently marketing.
We recently tested the software in-house and compared it with our own
in-house development which we have developed over the past few
years. We chose not to purchase the software but invest the significant
cost associated with the package into further development of our in-
house package. Even though the drawing package associated with the
CSEARCH is vary friendly, we found that some of the shifts calculated
by the package were poor in comparison with our own package.
e) Sadtler also sells a similar search program, called CSEARCH,
but it is much more expensive than the IBM database program we
purchased from them. We use the Sadtler data base for 13C at ---.
It has 34000 references and a relatively easy program for creating
structures. I use it mostly to find substructures and it has been
invaluable in confirming and/or suggesting structures of product
impurities, etc. Since we have the hardcopy versions of both the 1H
and 13C Sadtler spectra, I use the structure search to find the 13C
spectral number and then locate the 1H spectrum in our microfilm files.
The local salesman is bob petitti at (708) 582-0205.
f) Sadtler sell 13C databases and search software. One package is
PC based and can display 13C 'stick' spectra and structures and is easy
to use.Sadtler also supply CSEARCH which, among other things, can
simulate a 13C spectrum from a proposed structure. One can also
generate databases of your own data of, in principle, any nucleus for
incorporation into the package.
g) We recently completed a somewhat thorough evaluation of
Sadtler's C-Search software. In addition to being what I'll call a
"straight search" NMR database program, C-Search has spectral
estimation capabilities. It uses the approach of assigning H.O.S.E.
codes as a means to establish a relationship between structure and C-13
chemical shifts, as discussed by W. Bremser in Analytical Chemica
Acta, (1978). The advantage of this approach is that one does not have
to match the spectrum of an unknown with an entry in the database.
For example, an organic chemist may synthesize a brand new
compound and try to assign its C-13 spectrum. Though there will not
be a compound in the Sadtler library whose spectrum will match, there
may be enough compounds with similar sub-structures that closely
resemble those found in the new compound. The chemist may page
through, or better yet, use a computer to look up relevant examples.
Still, even with a computer, this may be tedious. C-Search serves as an
intelligent assistant to the chemist who wants to assign C-13 spectra of
novel compounds, though it also functions very well to help interpret
spectra of known compounds. The chemist might typically draw the
suspected structure of the compound using C-Search's molecular
editor. Next comes the "straight search" through the database for an
"exact" match. Usually none is found (if your problem is at all
interesting), so the chemist launches the predictive engine. The system
seeks relevant examples and uses their chemical shift data in
combination to produce a proposed spectrum. The key word is
"relevant" -- the estimated chemical shift data is only as good as the
quality of the database. If the database did not have spectra of
compounds with nitrosamine functional groups for example, then it
may not produce accurate shift estimates for the carbons in proximity
to the nitrosamine group of an unknown. C-Search also has other
capabilities, such as subspectrum estimation - e.g. if one is only
interested in aromatic carbon chemical shift estimates. DEPT
multiplicities, coupling constant data and carbon-carbon connectivity
information can be included with chemical shift data. The molecular
editor is very slick and there is a lot of ancillary information that you
can enter with the spectral data, such as physical properties and journal
references. Soon to come are capabilities to handle mixtures and
structural isomers. Later to come will be intelligent links into mass
spec and IR databases, which would lead to the system being a total
structure elucidation assistant. So did we buy C-Search? Not yet, but
we are very tempted. We will definitely take another look at future
versions. The cost is not trivial and we are waiting for more
capabilities as well as a little more ease-of-use. Others among us seem
to think that C-Search goes beyond our needs in some respects, while
in other areas we need more capabilities. We may go with Sadtler's
"straight search" program called SearchMaster. C-Search is already a
very fine product - for us it's just a matter of meeting our needs.
Footnotes: C-Search runs at least on Sun SPARC and Silicon Graphics
Indigo. We demoed it on a relatively outmoded Sun 4/110 and
performance was fine. It uses an X windows for display, so it can be
run to display over a TCP/IP network on PC's, Mac's or any other
workstation running X windows display software. SearchMaster runs
on Intel, MS-DOS, PC's running Windows. One could in principle run
SearchMaster on a UNIX workstation if it runs a product like Insignia
Solution's SoftPC emulation software. Our contacts at Biorad Sadtler
are Dr. Bruce Woods, who is an expert on C-Search theory and
practice; and Robert Petitti, the sales rep. for the midwest region - he
can provide further information of course, and can be contacted at 708-
582-0205. The number for Sadtler in Philadelphia is 215-382-7800.
h) I have recently performed a thorough testing of CSEARCH v. 5
Without entering in details, I can underline three main points:
1) This kind of software gives its best performance when used to
create a proprietary database of C13 spectra of closely related
structures to be used together with at least one of the two main
databases offered with the retrieval software. In this case, a high
increase in answer accuracy is observed already with a small number of
additional entries.
2) Since diastereoisomers, especially of rigid (e.g. polycyclic)
structures, can easily have C13 chemical shifts differing of ten ppm or
more, the lack of stereospecific assignments and structure drawing is a
major drawback. As far as I know, no one of the currently available
C13 database softwares has yet overcome this problem.
3) If no stereochemical difference is encountered, the goodness of
the fit depends (of course) on how similar a structure can be found in
the databases. The principle " same substructure, same subspectrum "
is well demonstrated in practice.
i) Sadtler also has a good database you can buy for your pc. There
you pay the $'s up front (several 10's of thousands, depending on how
many spectra you buy) and use if free after your initial purchase. I have
names of contact people if you want them. Both databases gave
assignments, structures, etc. I think they both give references also.
2) STN PRODUCTS
a) One of the most comprehensive is ChemInfo that supplied by
STN. This was formally the C-13 DataBase compiled by Bremser, et al
at BASF. Before STN "improved" it by setting it up for STN express
you could use almost any communications program to do searches or
spectral estimations based on HOSE codes. I used it for 7 years and
found it useful. A colleague and I tried, unsuccessfully to access it from
his Mac with STN express. After that I decided to return to using
microfiche.
Verlag did put out a set of microfiches based on the same database
above. The big disadvantages are two. First, the fiches are not updated
regularly. I don't even know if the current version is in print. Second,
because its in fiche form spectral estimation is, at best tedious. It is a
good source of assigned model compounds. Our most recent version is:
Fourth Edition, Carbon-13 NMR Spectral Data, A "Living" COM-
Microfiche Collection, W. Bremser, et al, VCH Publishers.
b) stn international 800-848-6538 ext 3731 or 614-421-3600 has a
bunch of data bases, one of which is c13-nmr/ir spectra online. they
seem to like pc's and macs and its done over the phone lines (again no
internet access..yet?) you send them an agreement and order form and
they give you a password ($25.00 to start an account and then connect
time).
c) I have also used the STN 13C database available from CAS.
This is an even larger data base and has the advantage that you can
input a structure and the program will suggest the chemical shifts based
on those references with similar carbon atoms. The STN search uses a
teletype like interface and thus does not draw structures as readily as
the Sadtler program and I found considerably less user friendly, but
may be a good way for you to start since the charges are based on each
search.
d) STN International has much larger NMR database. As I
remember, they started about six years ago and now are collaborating
with Chemical Abstract. All you need is a terminal with keyboard and
printer (or screen display) hooked up with a modem to have access to
their database. I think they've send their flyersto every ENC. Try to
empty your ENC bags to see if you can find them (I could not find
mine). I have their old catalog, which lists their address as follows:
STN-Columbus
c/o Chemical Abstract Service
2540 Olentangy River Road
P.O. Box 02228
Columbus, OH 43202 USA
Tel: 800-848-6533 (in USA except Ohio)
800-848-6538 (Ohio and Canada)
614-421-3698 (worldwide)
Fax: 614-646-3411
e) One possibility that you might want to check out is the online
SPECINFO database through STN International. I have not used this
file, but I am quite impressed with what I have used. The SPECINFO
database contains 80,000 NMR spectra. The info I have states that
several data generation packages including CHESS, COUPCAL,
SPECAL, EDSPEC, and GETSPEC, can be used within SPECINFO.
These packages can be used to search for chemical structures, estimate
coupling constants, estimate spectroscopic information from a query
structure, and identify compounds with matching or similar spectra. If
you do not have an account with STN, they can be reached at 800-848-
6533.
f) Specinfo has databases for other nuclei eg 19F, 17O , 15N , and
especially 13C (and probably others as well).
Specinfo databases can be purchased - either complete or as smaller
pieces by chemical type eg polymers or steroids etc.
In Europe we can access STN/Online which is essentially the
Specinfo databases and do searches of structures or functional groups
etc. The problem here is that one's own data cannot be entered without
a deal of effort.This can be expensive -not only in 'phone calls!
Specinfo is handled (in Europe) by :-
Chemical Concepts
PO Box 10 02 02
D-6940 Weinheim
Germany Fax (+49) 62 01 60 64 30
g) we use SPECINFO, a 13C/IR/MASS library containing some
100,000 13c spectra. Available as in house version or on line via STN
international. Vendor is Chemical Concept, Weinheim, Germany. We
are very pleased with the performance of this data system.
h) I just had a short course using SPECINFO, its part of the STN
(Chem abstracts) database. They have 90,000+ compounds with a
variety of info, most have 13C, some have other nuclei and/or IR and
mass spec data. They have lots of options for searching, you can draw
structures with STNExpress software and upload then. Its been fairly
useful when I've used it, the online charges are pretty big, its easy to
run on a few hundred bucks, I don't know what the academic rates
are......
i) Bruker used to supply a 13C database package. I think that they
do not do that any more and the data is the basis of the specinfo
system. Bruker also supplied the BASIS database system for ASPECT
2000/3000 computers. This is a flexible piece of software but no data is
supplied and entering data and structure drawing is not trivial. Perhaps
the 13C data could be incorporated?
j) The cheapest database though at the moment not on computer is
the Aldrich Library of 13C and 1H FT NMR spectra. 12 000 1H
300MHz + 12 000 13C 75MHz spectra in 3 volumes for 600 pounds.
PROTON
1) For proton NMR data an excellent book source is the Handbook
of Proton NMR Spectra and Data.This is an approximately $1000 set
of five books from the Asahi Research center in Japan. Each book
contains about 800 interpreted 90MHz spectra and puts the Aldrich
Library of NMR spectra to shame. There are also tables in the back that
allow you to look up the PMR shift of a proton in a specific
environment.
The reference for this set is:
Handbook of Proton-NMR Spectra and Data, Vols 1-5, Edited Asahi
Research Center Co, Ltd, Tokyo Japan, Academic Press, Inc, 1985
2) As far as we know there is, at present no satisfactory 1H spectral
database - SPECINFO in Germany are working on one but the
difficulty of handling chemical shifts and coupling constants and
different solvent systems is the main problem here.