ScienceSoft, LLC

CHAPTER 16: Report Workwindow

This chapter describes the Report workwindow summarizing the automated two- and three-dimensional spectral analysis results. Its input fields and resulting program output are explained. (To display all the input fields available in this workwindow, select from the Edit menu [Preferences...]. Set Mode: [Full NMRanalyst], [Show All Input Fields] switches, and click [OK].) See CHAPTER 12: "Using the Workwindows" for a general description of the function and use of a workwindow.

Program report, controlled by the Report workwindow, assembles the information obtained from the 1D Analysis, FFT, and nD Analysis workwindows for the selected multidimensional spectrum type. The analyze program, controlled by the nD Analysis workwindow, examines one fitting area at a time and does not attempt to assemble the analysis results from different fitting areas. The program report summarizes this information and provides warning messages in case of ambiguous signal assignments, a phase mismatch of a reported correlation with the overall phase functions, or overlapping correlation areas.

The AssembleIt workwindow can combine the Report workwindow results from different spectrum types. Its FindIt component identifies the best matching molecular structures for the observed NMR data. Its AssembleIt component derives the most likely skeletons from this often incomplete and inconsistent correlation information, see CHAPTER 17: "AssembleIt Workwindow" for details.

16.1 Input File

The label of the input field reflects the dimensionality of the selected spectrum type. Enter in this input field the file name specified in the nD Analysis workwindow Determined ?D Correlations field. The file contains the numerical description of identified patterns in the analyzed multidimensional spectrum. (Key: Fn2D)

16.2 Output Files

The function of the Report workwindow is to summarize the determined molecular structural information. The graph of determined correlations is saved when a file name is provided in this input field. The graph can be displayed using the Graphic workwindow. See CHAPTER 18: "Graphic Workwindow".

For 2D INADEQUATE spectra, the resulting graph shows the determined carbon skeleton. For heteronuclear spectra, the displayed fragments consist of one carbon atom with correlated proton frequencies. For NOESY, ROESY, and TOCSY spectra, each displayed fragment contains a proton surrounded by correlated proton frequencies. The AssembleIt workwindow allows combining the information from different spectrum types and deriving the possible carbon skeleton(s) from it. (Key: FnSklt)

Specify a name for the plot file of determined phase values and phase functions. This plot shows the precision with which the report program determines the spectral Phase Functions. The plot file is created when a file name is specified in this field. Use the Graphic workwindow to display it. (Key: FnPse)

The Report workwindow produces a connectivity table and warnings. This output file contains a plot of areas of 2D or volumes of 3D spectrum showing the location of determined correlation signals. The plot file is created when a file name is specified in this input field. Use the Graphic workwindow to display it. (Key: FnArea)

The normal (Gaussian) distribution of parameter precisions or agreements of non-correlation patterns allows calculating correlation detection probabilities. Specify a name in this input field to save a plot file of the distribution of analyzed correlation areas, the threshold for detected correlations, and the calculated distribution function. Use the Graphic workwindow to display it. (Key: FnIP)

The Precision Distribution, described above, allows effective distinction of signals from random noise. But spectral distortions, especially T1 ridges, are not random and often consist of signals which match the spin system model. Specify a filename in this input field to save the integral distribution and use the Graphic workwindow to display it. NMRanalyst determined integrals correspond to the integrals of well-phased signals. Based on this integral distribution plot, choose an integral value threshold above the noise and ridge signal responses. To exclude the weak signals, specify this threshold in the nD Analysis workwindow Thresholds: Integral field as described in CHAPTER 15: "nD Analysis Workwindow" and rerun the analysis. (Key: FnInt)

Acquiring a 1D carbon spectrum is 5700 times less sensitive than a 1D proton spectrum. A 1D nitrogen spectrum is 50 times worse than carbon. Acquisition of a 1D carbon spectrum often takes as much time as a multiplicity edited HSQC, HMBC, and DQF-COSY spectra combined. The acquisition of a nitrogen spectrum is impractical. Equivalent resonance information can be derived from HSQC and HMBC type spectra. If a file name is specified, the F1 resonance list is redetermined and saved. From HSQC, only protonated resonances can be obtained. An HMBC is often not phase sensitive in F1. So use as F1 resolution no less than the distance between the closest heteroatoms to be resolved. This field is displayed for the HMBC, N15_HMBC, HSQC, COUPLED_HSQC, and N15_HSQC spectrum types. (Key: FnResF1)

16.3 Select Items to Be Reported

When the [Everything] option is selected, all information determined for every correlation signal is displayed:

 All CORRELATION Information
 #    6 C 12 C 13  CORRELATION:  Fa= 30.038    Fb= 22.013    J= 31.97          
 Ia1,2  65.8241      555.824     -1.000      65.8241      555.824      67.77   
 Ib1,2  65.8241      585.838     -1.000      65.8241      585.838      65.30   
 Fa,Fb  30.0482      30.0382      .2637E-02  22.0183      22.0129      .2362E-0
 Pa,Pb -2.21128     -1.98546      .1980     -2.24784     -2.08620      .1836   
 Ta,Tb  .200000      .200000     -1.000      .221400      .221400     -1.000   
 Ts,J   .233319      .233319     -1.000      35.0000      31.9676      .2669   
 Fd,Pd  52.0665      52.0511     -1.000     -.149412     -.197273      .8358E-0
 Tr,Td  .100000      .100000     -1.000      .109368E-01  .109368E-01 -1.000   
 iter   8   Agree  .1416     ChiSq  .101814574E-01 

The information about a correlation labeled #6 is shown. The first line (#6 C12 C13 CORRELATION: Fa=30.038 Fb=22.013 J=31.97) is the summary line discussed below. The following lines, from top to bottom, contain the parameters for both spins of the reported correlation:

Both integrals for spin A
Both integrals for spin B
F2 transition frequency of spins A and B
F2 phase of spins A and B
Effective relaxation time of spins A and B
F2 acquisition time and coupling constant
F1 transition frequency and phase
F1 relaxation and acquisition time

The six columns of numbers in this table specify, from left to right, the initial estimate, the final value, and the error value for the first parameter, and the initial, final, and error value for the second parameter. A negative error value indicates that the corresponding initial parameter estimate was not optimized during the regression analysis.

EXAMPLE: In the above parameter list, the estimated coupling constant (J) for this bond in a 2D INADEQUATE spectrum is 35.0000 Hz, and the best-fit determined coupling constant is 31.9676 Hz with a marginal standard deviation (uncertainty) of ±0.2669 Hz.

The last line in the above parameter list gives some statistical information about the reported pattern. Of particular interest is the agreement value (Agree). It specifies which fraction of the signal power in the experimental correlation area can be explained by the reported pattern. A perfect fit (no noise) would give an agreement value of one. The reported agreement value of 0.1416 (14%) indicates a strong correlation signal given the detection limit of NMRanalyst.

When the [Summary Lines] option is selected from the Report About Correlations option menu, the following type of list is generated:

 CORRELATION Summary Lines
 #    6 C 12 C 13  CORRELATION:  Fa= 30.038    Fb= 22.013    J= 31.97
 #   16 C  3 C 13  CORRELATION:  Fa= 48.792    Fb= 22.016    J= 33.62
 #   19 C  2 C  3  CORRELATION:  Fa= 54.845    Fb= 48.790    J= 32.66
 #   31 C  3 C  4  CORRELATION:  Fa= 48.794    Fb= 45.381    J= 29.13
 #   32 C  4 C 14  CORRELATION:  Fa= 45.383    Fb= 19.644    J= 37.37
 #   34 C  4 C 15  CORRELATION:  Fa= 45.384    Fb= 19.374    J= 36.93

Each line describes one identified correlation signal and lists the correlation number, the number of each of the correlated 1D resonances,1 both chemical shifts, and for DQF-COSY, HMBC, N15_HMBC, COUPLED_HSQC, and INADEQUATE spectra the coupling constant. For other spectrum types, the average integral instead of the coupling constant is listed.

For the [Nothing] menu item, neither the summary lines nor the numerical values for detected correlations are reported. (Key: RBond)

The items of this option menu correspond to the Report About Correlations option menu above. Non-Correlations typically are not of interest. So, the [Nothing] option is selected by default. However, if an expected correlation is not reported, the corresponding summary line can be inspected to see why program analyze did not identify this pattern as a correlation:

 Non-CORRELATION Summary Lines
 #    1 C  7 C  9  No correlation detected: No pattern found.
 #    2 C  6 C 10  No correlation detected: dI4 Fa
 #    3 C  8 C  9  No correlation detected: dI4 Fa
 #    4 C  7 C  8  No correlation detected: dI1 dI4
 #    5 C  5 C 11  No correlation detected: dI1 dI4 Fa Fb
 #    7 C  3 C 14  No correlation detected: dI1 dI4

The best-fit pattern found for fitting area #1 was not identified as a correlation because at least one of the determined resonance frequencies lies outside the fitting area. The best-fit pattern found in fitting region #5 was not reported as a correlation because the determined parameter precisions for the A doublet (dI1) and the B doublet (dI4) are not sufficiently high and both determined transition frequencies are significantly different from the 1D resonance frequencies (Fa, Fb). For a more detailed description, see the explanation of the Detection switches in CHAPTER 15: "nD Analysis Workwindow". The determined numerical values for all non-correlations can be displayed by selecting the [Everything] option. (Key: RNBond)

The Report workwindow shows a table assigning all multidimensional spectral resonances to the 1D resonances for each spectral direction. For heteronuclear spectra, one table is displayed listing the correlations to the F1 resonances, one listing the correlations to the F2 resonances, etc. For each correlation, the respective chemical shift, coupling constant (or average detected integral), and correlation index number are given:

 Shift | Shift  J  Index | Shift  J  Index | Shift  J  Index |...
 [ppm] | [ppm] [Hz]  /   | [ppm] [Hz]  /   | [ppm] [Hz]  /   |...
  56.86| 45.39 31.8    36| 30.05 28.6    52|  9.32 40.9    44|

This 2D INADEQUATE Connectivity Table fragment specifies that the 1D carbon resonance at 56.86 ppm is correlated to three carbons resonating in the 1D spectrum at 45.39, 30.05, and 9.32 ppm. The table also shows the corresponding coupling constants (31.8, 28.6, and 40.9 Hz) and the correlation index numbers (#36, #52, and #44). The index numbers allow locating further information about the correlations as described for the Report About Correlations option menu.

Two-dimensional spectra can give rise to ambiguous correlation signal assignments. Ambiguities are reported by the report program in the Connectivity Table when the [Ambiguous Signal Assignment] switch is selected. All ambiguous interpretations are listed with additional messages:

 Shift | Shift  J  Index | Shift  J  Index | Shift  J  Index |...
 [ppm] | [ppm] [Hz]  /   | [ppm] [Hz]  /   | [ppm] [Hz]  /   |...
  16.77| 50.48  35.5  105|                 |                 |
          Ambiguity index# 105 and index#(s) 288
  16.79| 50.48  35.5  288|                 |                 |
          Ambiguity index# 288 and index#(s) 105

An ambiguity is reported when the same correlation signal is found while examining two overlapping fitting areas. In the 2D INADEQUATE table shown, both ambiguous regions involve the resonance at 50.48 ppm, which is correlated (bonded) to either the carbon resonating at 16.77 ppm (#105) or that at 16.79 ppm (#288). The three Hz F1 shift difference between the two resonances (500 MHz spectrometer) is smaller than possible isotope shifts between 1D and 2D carbon resonances, causing both assignments to be indistinguishable. When an ambiguity is encountered, information from other sources is needed to uniquely assign the involved correlations (carbon-carbon bonds). This ambiguity might be a disappointing discovery to the spectroscopist. However, one of the major strengths of NMRanalyst is to reliably detect and report such ambiguities.

The [Redetermine 1D Chemical Shifts] switch is mutually exclusive with this [Ambiguous Signal Assignment] switch, because it removes all but the most probable assignment from a collection of mutually ambiguous assignments. Selecting the [Ambiguous Signal Assignment] switch displays ambiguous correlations in a different linestyle in the NMRgraph correlation editor. See CHAPTER 19: "NMRgraph: Molecular Correlation Editor" for details. (Key: LAmbig)

When the [Improper Phase] switch is selected, those correlations with phase values inconsistent2 with the determined phase functions are identified in the Connectivity Table:

 Shift | Shift  J  Index | Shift  J  Index | Shift  J  Index |...
 [ppm] | [ppm] [Hz]  /   | [ppm] [Hz]  /   | [ppm] [Hz]  /   |...
 135.50|132.63  60.7   31|128.54  65.4   12|                 |
        Phase mismatch index#(s)    31    12

An individual phase value determined from a 1D spectrum can be shifted by 2 (360 degrees) and two phase values from a single transition determined from a multidimensional spectrum can be shifted by (180 degrees) in two orthogonal directions without changing their appearance. For example, for 2D INADEQUATE spectra, the bond signal model has three phase values (one F1 and two F2 phases). A phase mismatch occurs when one phase is shifted an odd and the other phases an even number of times, or the other way around. The shifted phases and determined phase functions are saved in a plot file as described earlier in the Output Files section, and the plot can be inspected using the Graphic workwindow.

Even if the [Correlation Table(s) With Warnings on:] switch is not selected, selecting the [Improper Phase] switch displays such correlations in a different linestyle in the NMRgraph correlation editor. See CHAPTER 19: "NMRgraph: Molecular Correlation Editor" for details. (Key: LPseMm)

When two correlation patterns identified in different fitting areas are not identical (which would indicate an ambiguity) but both fitting areas partly overlap, the report program issues an overlap warning if the [Fitting Area Overlap] switch is selected:

 Overlap windows # 105 - # 288 with transitions (A,A) (B,B)

NMRanalyst was designed to distinguish expected spin systems from noise. Overlapping correlation signals can neither be described as isolated signals nor as random noise, and a severe overlap of two or more correlation patterns can possibly lead to a false positive correlation test. The best approach is to acquire the spectrum with sufficient resolution to avoid such problems. If this is not feasible, the overlap warnings allow quickly identifying the spectral regions that could cause misinterpretations of the data.

Selecting the [Fitting Area Overlap] switch displays such overlapping correlations in a different linestyle in the NMRgraph correlation editor. See CHAPTER 19: "NMRgraph: Molecular Correlation Editor" for details. (Key: LOvrlp)

Without selection of this switch, multidimensional resonances are assigned to 1D resonances and the 1D resonance values are shown in the displayed correlation tables and related structure plots. If this switch is selected, new lists of 1D resonances are determined from the multidimensional data. Then the correlation tables and structure plots show resonance frequencies directly determined from the multidimensional spectrum. Selecting this switch removes ambiguous assignments from the correlation tables except for the most likely interpretation. It is recommended to not select this switch if the 1D information was determined from an actual spectrum and to select it if generic 1D information or spectral projections were used for the analysis. For carbon and nitrogen HSQC and HMBC spectra this switch is not shown. Chemical shifts are redetermined if a Redetermined F1 Resonance List value is specified. (Key: LSimpl)

16.4 Generate NOE Build-up Curves

The panel for the generation of NOE build-up curves is displayed for NOESY and ROESY spectrum types. The shown switch selects the generation of NOE build-up curves. NMRanalyst specializes on the automated analysis of experimental multidimensional NMR spectra. This NOE build-up curve extension provides a simple but powerful way to combine information from a series of NMR spectra. The derived build-up curve slopes are more reliable for the determination of the 3D structure of molecules than the individual signal volumes. (Key: BuildUp)

Up to five NOE spectra acquired with different mixing times can be incorporated in the build-up curves. The mixing times entered in these input fields should be specified in milli-seconds, and the software accepts values between zero and 60,000 ms (one minute). The spectrum file name (input fields described below) and the corresponding mixing time need to be specified in the same line of the table. The order in which the spectra are listed in the table does not matter. (Key: MixT1, ..., MixT5)

The build-up curve generation starts with the analysis results of the corresponding component spectra. It is recommended to carefully analyze the spectrum of the highest mixing time since this spectrum will have the highest signal-to-noise (S/N) ratio. Only the identified correlations need then be analyzed in the spectra acquired with shorter mixing times. Finally, all nD Analysis workwindow output files of the component spectra should be entered in these input fields and the corresponding mixing times should be entered in the Mixing Times [ms] fields described above. (Key: FnNOE1, ..., FnNOE5)

Use this input field to specify a name for the created plot file. If no name is specified, build-up$$.plot is used. The two dollar signs in the default name expand to a unique process-ID at run-time. The build-up curve plot is created when the Report workwindow is run with the [GENERATE NOE BUILD-UP CURVES] switch selected and the Mixing Time [ms] and 2D Analysis Output File For NOE Spectrum input fields contain valid entries. The numerical analysis results are displayed from the Report workwindow output screen and the created plot can be displayed by the Graphic workwindow. (Key: FnBuild)

NMRanalyst provides the user interface for this build-up curve generation. The report program calls the corr2plot shell script for the actual curve generation. Script corr2plot eliminates curves with negative slope (likely diagonal signals). When three or more mixing time spectra are given, it eliminates those build-up curves containing only one determined volume (possibly a spectral artifact). Each build-up curve shows all determined volumes with error values and the best linear interpolation of these weighted points. The corr2plot script is a simple text file in Bourne shell syntax. It can be customized as desired.

1The numbers are (if not modified by the user) the position of the 1D resonances in the (decreasing frequency sorted) 1D spectral line list(s).

2NMRanalyst uses an improved version of the algorithm given in Dunkel, R. U.S. Patent 5,218,299 for resolving the 2 ambiguities in 1D and ambiguities in determined multidimensional phase values.

ScienceSoft, LLC