Assemble 2.0 Tutorial

Brief Description | First Example | Fundamental difference between non-overlapping fragments and substructure constraints | Atom Tags | Assemble as a tool in structure elucidation | Ranking | Postprocessing

Ranking according to nmr spectral information

Assemble accepts structural, not spectral information. The interpretation task is the responsibility of the user. It does not mean, that all the interpretation is done manually. There are tools available to automatically transform spectral into structural information. Two of them have been connected to Assemble. Such interpretation aids can be of great value, but their strength also bears a certain danger of misinterpretation. There is almost always a certain amount of doubt what a spectral feature means in terms of connectivity between atoms. Occasionally also experts make mistakes. Automatic procedures are no more reliable. The user is still responsible for the correctness of the predictions. Use the tools with care.

The two interpretation modules that come with Assemble estimate the nmr chemical shifts of the candidates generated. The values are then compared to the experimental shifts. The structures are ranked according to a criterion of agreement between estimated and experimental figures. Experience shows that the correct structure is often near the top of the list, even if only the molecular formula and the most reliable data is provided.


Ranking according to the C-13 nmr spectrum

The C-13 nmr spectral information is given as the chemical shift of the signals and their multiplicity in the off-resonance decoupled spectrum. This is the information about the number of immediately bonded hydrogen atoms. This information is also available utilizing various multi-pulse nmr experiments such as DEPT.

Try the module by using the structure elucidation example, this time providing only the molecular formula and the C-13 nmr data. The C-13 nmr spectrum shows 6 signals. Their multiplicity is also provided. The letters D and T below the chemical shift mean doublet (the trace of a CH group) and triplet (a CH2 group).

First run Assemble, providing only the molecular formula. If you have a current project running, delete all input fields. Alternatively you can open a new project. Specify the molecular formula as C6H10O2. Run the assembler. There are 4869 structures generated.

From the "Project" pull-down menu select "Rank Output".



A window pops open. Choose the "C-NMR" tab from the top of the window.



If the multiplicity information is available, as in the current example, choose the "Off-Resonance" method.



The entire spectrum in jcamp format can be analyzed to provide the shift information. Alternatively the numerical values can be entered manually. The format is shown when the "Set Example" button is clicked.



Now you can type in your own data. For convenience, the list of shifts and multiplicities is given here. Move the data with copy/paste to the input window.

135.0	1	CH
116.8	1	CH2
72.2	1	CH2
71.1	1	CH2
50.8	1	CH
44.0	1	CH2

Depending on the type of your hardware, the calculation may take a few seconds up to several minutes. The result is shown in a window very similar to the ordinary Assemble output window.



As you see, the correct structure ranks first with a mean deviation of 1.20 ppm.

The designation 6/6/6 means that 6 shifts (all of them) have been mapped to carbon atoms. The structure is predicted to have 6 signals in the C-13 nmr spectrum (the second number) and finally the third "6" specifies the number of signals you just specified as experimental shifts.

The maximum deviation amounts to 2.30 ppm. This number is very moderate. It is typical for a correct structure.

When these figures are compared to those of the structure ranked second, it becomes apparent that the wrong structure, and all others ranking below, show much larger deviations. This is a strong hint to accept the first structure as being the correct solution.


Ranking according to the H-1 nmr spectrum

Switch to the "H NMR" method and click the "Set Example" button as you did before.



The H-1 nmr spectral information is given as the chemical shift of the signals and their integral. A coupling pattern may be specified in addition. If you don't want to specify a coupling pattern, enter the letter "m" (multiplet). Here is the shift information from the experimental spectrum ready to be copied.

5.9	1	m
5.3	2	m
4.1	2	m
3.7	1	m
3.4	1	m
3.2	1	m
2.8	1	m
2.6	1	m

Start the calculation by clicking the "Calculate" button.



The window popping up shows the ranking results.



The structures look strange. In addition the text in the output fields, written in red, says "Error: Shift prediction not possible". What happened? When you look at the first field, you also see "0/10/10". This means that not a single shift has been mapped to any hydrogen atom in the structure. All values are outside the acceptable shift interval for the particular combinations of chemical environment and signal integral. This indicates a very unprobable candidate. Why does it rank first?

It is a measure of security. Sometimes the shifts of the correct solution cannot be estimated adequately due to insufficient information about its chemical environments. If this happens, you are very likely to miss the correct solution. This is the worst possible outcome. Therefore these structures are shown first, so you are warned.

The 7th structure is the first one for which no error is reported. Two of the experimental shifts are mapped (ore one shift with integral 2). As there is ample choice for so few mappings, the mean and maximum deviations tend to be small. Therefore these structures rank very high, though the bad mapping again shows the structure to be an improbable candidate.

When you look at the next set of structures, ranking 10th to 18th, you see the first structure to be the correct solution. The results are shown as text in blue color. The mean and maximum deviations are 0.13 ppm and 0.34 ppm respectively. All the shifts have been mapped to the hydrogens of the molecule.



The color of the text gives you a hint about the quality of the spectrum estimation.

red	Estimation of at least one chemical shift was not possible as the information about the chemical environment is not available. The corresponding shift is not used in comparison with the experimental spectrum. Keep in mind that generated candidates may look weird. Very possibly this is the case if a red entry is encountered.
magenta	The estimation of at least one shift was not possible as the corresponding chemical environment leads to high fluctuations naturally. This is the case for hydrogen bonded to hetero atoms. If some less vital information about the chemical environment is missing, the estimation is of lower quality but still used for comparison with the experimental spectrum.
blue	All chemical shifts could be estimated satisfactorily.

To see the next entry written in blue, you have to step forward to the structure ranking 58th.



The mean and maximum deviations are still quite reasonable, which is also the case for the next few entries with all shifts mapped. Therefore the result of the ranking is not as striking as in the former example using the C-13 nmr spectrum.

If you don't want the structures without predicted shifts to be reported at the beginning, you can request they appear at the end of the list. Select the radio button next to "at the end of ranking result" as shown.



Repeat the calculation. The correct structure now ranks 5th.