ScienceSoft, LLC
www.ScienceSoft.net
TOC PREV NEXT INDEX


CHAPTER 7: Tutorial V: Additional NMRanalyst Features

The previous four tutorials cover the NMRanalyst analysis of 1D and 2D spectra, the FindIt structure identification, and the AssembleIt structure elucidation. This tutorial adds determining molecular structures from an HSQC FID (SECTION 7.1: "Using an HSQC FID for the findit Structure Identification"), generating simulated spectra (SECTION 7.2: "Displaying Experimental, Simulated, and Residual NMR Spectra") and 2D surface plots (SECTION 7.3: "Surface Plot of a 2D Spectrum"). It also covers analyzing 3D data (SECTION 7.4: "Analyzing 3D Spectra"), incorporating NMRanalyst programs into shell scripts, macros, and other software systems (SECTION 7.5: "Incorporating NMRanalyst Programs in Other Software"), and the online help system (SECTION 7.6: "The On-line Help System"). In NMRanalyst, select [Edit] [Preferences...], set Mode: to [Full NMRanalyst], and select the [Show All Input Fields] switch, as this tutorial describes less commonly used features of the software. Click the [OK] button.

7.1 Using an HSQC FID for the findit Structure Identification

SECTION 3.1: "findit Script for Molecular Structure Identification" bases the structure identification on a 1D carbon FID. findit can substitute a missing proton or carbon FID by the corresponding F1 and/or F2 projection of an HSQC or HMQC spectrum. SECTION 6.1: "1D Carbon, 1D Proton, Edited HSQC, and HMBC Analyses" shows the strychnine HSQC spectrum and its F1 and F2 projections. The 2D spectrum phasing does not matter for absolute-value projections. findit converts both projections to NMRanalyst format. It then determines best matching structures.

Copy the strychnine directory from the Tutorial IV to a temporary location. Delete its proton and carbon FIDs to keep findit from using them. From the NMRanalyst UNIX Shell window, execute the following commands to run findit:

 % cp -r E:/strychnine C:/tmp
 % cd C:/tmp/strychnine
 % rm -r proton.fid carbon.fid
 % findit
 ##########################################################################
 #  "findit [Formula_Constraints]"              (C) 2010 ScienceSoft LLC  #
 #  Best matching of 14.5 million molecular structures for 1D H&C FIDs,   #
 #  uncompressed JCAMP .dx files, proton & carbon line lists, HSQC, or    #
 #  "Formula_Constraints". Use NMRanalyst to inspect intermediate steps.  #
 ##########################################################################
 C:/tmp/strychnine
 Varian HSQC: C13_ghsqc.fid/procpar 
 Transforming HSQC projection(s).
 Determining most likely FindIt structures...
    1: 0.964780 (    5304)    2: 0.942377 (  273117)    3: 0.929914 (16348068)
    4: 0.929087 (16373846)    5: 0.926625 (10645330)    6: 0.925683 (17556896)
    7: 0.924923 (21657150)    8: 0.923669 (16301457)    9: 0.922883 (11489638)
   10: 0.922523 (17408781)

The top structure (PubChem CID 5304) is the correct strychnine structure. Replacing the proton and carbon spectra by HSQC projections leads to some compromises. Integrals of a proton spectrum identify likely -CH3 and -OCH3 groups. For the proton projection, only the signal heights are known, but not their line shapes and derived integrals. With regard to the carbon projection, only carbons with attached protons are detected by HSQC. So findit sets in the AssembleIt workwindow the F1 projection in the 1D: DEPT rather than in the 1D: Carbon input field. HMQC spectra give worse F1 line shapes than HSQC projections. But the HSQC or HMQC spectra remain faster to acquire than a 1D carbon spectrum. Overall, proton and carbon 1D FIDs are preferable for structure identification over 2D spectrum projections.

7.2 Displaying Experimental, Simulated, and Residual NMR Spectra

For spectral analyses, the Fourier transformed spectrum is created. Phase functions are determined and locked when feasible. The Report workwindow saves the location of identified correlations. Combining this information in a phased experimental spectrum with detected correlation locations and 1D spectral projections is the default way to visualize these results. SECTION 4.7: "Prednisone Spectrum With 1D Carbon & Correlation Locations" gives an example. For the detailed visual inspection of a correlation area, the experimental, simulated, and residual spectral areas can be created and displayed as described in SECTION 4.8: "Experimental, Simulated, and Residual Spectrum of One Correlation". This section describes creating such plots for the complete spectrum.

Switch NMRanalyst to the strychnine directory created in the previous tutorial. To create a simulated HSQC spectrum, switch to the HSQC spectrum type and the Report workwindow output screen. In the middle of the report program output is:

 Correlation Summary:
 103 106 108 215 304 306 308 395-396 477 479 597 602 698-701 703 708 807 904-905 
 1002 1004 1021-1022 1200 1206-1207 1209 1326 1329 1413 1416 1487-1490 1498 1500 
 1582 1692 1697 1703-1704

Select the three lines of identified correlation index numbers. Click the right mouse button and select [Copy Selection]. Switch to the 2D Analysis workwindow. Click the Analyze Correlation Index #s field to select it and delete any previous entry. Click the right mouse button and choose [Paste]. Scroll to the bottom of the input screen, and select the [Add Correlations] switch. Deselect the [RUN REPORT WORKWINDOW AFTER SUCCESSFUL ANALYSIS] switch. Run the workwindow. Every best-fit correlation area is added to the simulated spectrum.

The shown three spectra display from top to bottom the experimental, simulated, and residual spectra. To create the upper one, delete in the Graphic workwindow the Add Corr. Locations, Add F1 1D Spectrum, and Add F2 1D Spectrum entries. Set File With Plot Data to hsqc.spec, Relative Size: Y to 0.33, Contour-Lines: Max to 0.03, and start the workwindow. To create the simulated spectrum, change File With Plot Data to hsqc.spec.sim and run the workwindow. The residual spectrum is the difference between the experimental and simulated spectra. In the Graphic workwindow set File With Plot Data to hsqc.spec, Subtract Spectrum to hsqc.spec.sim, and run the workwindow. To adjust the size of a plot in these windows, press the right mouse button in a plot window and move it up (increase) or down (decrease).

7.3 Surface Plot of a 2D Spectrum

As in the last section, switch to the strychnine dataset and set the spectrum type to HSQC. Scroll to the bottom of the Graphic workwindow input screen, set 2D Plot Type to [Surface] to render the surface rather than the default spectral contour plot. To display the experimental strychnine spectrum, set File With Plot Data to hsqc.spec and delete the Subtract Spectrum, Relative Size: Y, and Contour-Lines: Max entries. NMR resonances tend to be small compared to the screen resolution. Increase Plot Window: Size to 600. Start the workwindow to obtain the shown plot. It can be interactively positioned, rotated, and zoomed as described in CHAPTER 20: "NMRplot: Plotting Multidimensional Spectra".

7.4 Analyzing 3D Spectra

The oligomer_3D_SPECTRUM dataset contains the 3D HNCO FID of 0.5 mg of the AGAGAAAG oligomer (containing a sequence of amino acids Alanin and Glycin)1. NMRanalyst's 3D_SPECTRUM spectrum type analyzes all phase sensitive 3D spectra as though they were heteronuclear spectra. Create a copy of the provided oligomer_3D_SPECTRUM dataset. Start the Directory Editor, specify the oligomer_3D_SPECTRUM directory in the NMRDATA field, delete the NMRSPEC entry, and click [OK].

The 1D line lists for the 3D analysis are normally generated generically as described in SECTION 5.5: "Structure Identification Without 1D Carbon Spectrum" and SECTION 6.2: "Generation of a Generic 15N Resonance List". Other generation methods described for 2D spectra can be used as well. For this heteronuclear 3D dataset, the three line lists (files: C_lines, N_lines, H_lines) are provided. Select the FFT workwindow and load HNCO.fid/procpar. A 3D Varian procpar file does not indicate which decoupler frequency belongs to which spectral dimension. So follow NMRanalyst's mapping convention for the data acquisition as implemented in shell script varian2txt, or modify this shell script to fit your 3D acquisition preferences.

When the Transform Order: menu lists the pulse sequence used, such as [Varian hnco] in this case, select the corresponding item to request a native FFT. If the used pulse sequence is not listed, select the [Coefficient File] item and the 40 transform coefficients are read from the file specified in the 3D FFT Coefficient File field. The coef coefficient file for the current dataset is supplied and VNMR can create coefficient files for standard Varian pulse sequences. Choose one of these two native transforms now and perform it by clicking [Start].

To analyze the 3D HNCO spectrum, click the [3D Analysis] tab and click [Start]. The analysis of this 3D spectrum takes about one minute:

 Display of Correlation Information
 ==================================
 #     1 H  1 C  1 N  1  No correlation detected: dI1
 #     2 H  1 C  1 N  2  CORRELATION:  F1= 163.393  F2= 105.190  F3=   8.174
 #     3 H  1 C  1 N  3  CORRELATION:  F1= 163.391  F2= 105.190  F3=   8.174
 ...
 #   140 H  5 C  4 N  7  No correlation detected: dI1

To summarize the 3D analysis results, run the Report workwindow. Three phase functions (file phases.plot), 3D fitting volumes (file volumes.plot), integral (integrals.plot) and parameter precision distributions (precisions.plot), and a structure.plot file are saved equivalent to the 2D spectrum analyses. Display the shown structure.plot file from the Graphic workwindow. This oligomer has eight amino acids and seven HNCO signals are expected and observed. The carbonyl carbons (shifts in red representing the 4 carbon valences) are bonded to 15N atoms (shifts in green representing the 3 nitrogen valences) which are bonded to protons (shifts in orange representing one proton valence). Bond labels show the correlation index number under which further information can be found concerning the corresponding correlation signal. Display the P(I) distribution (file precisions.plot) for this dataset. NMRanalyst is significantly more sensitive than required for the analysis of this spectrum. As for 2D, the software is very beneficial for analyzing low S/N 3D datasets.

SECTION 4.8: "Experimental, Simulated, and Residual Spectrum of One Correlation" describes the creation of experimental, simulated, and residual surface plots of an analyzed 2D spectral fitting window. For 3D spectra, the equivalent plot shows the experimental, simulated, and residual 3D iso-surfaces. In the 3D Analysis workwindow, set the Analyze Correlation Index #s field to 49 (the second weakest correlation), select the [Combined Experimental, Simulated, Residual Spectrum] switch, and run the analysis of this correlation. From the Graphic workwindow display the created Fa162.92_Fb90.91_Fc7.89.plot file. Like other NMRplot displays, this plot can be resized, rotated, zoomed, and moved. The sinc character of 3D signals might cause them to not be equally visible in all eight fitting volume phase components. This does not compromise the computerized spectral analysis as all eight phase components are examined simultaneously. But to get the best display of a signal in a fitting volume, several phase components might need to be examined. The NMRplot window allows changing the display among the phase components through the [Component] menu.

HNCO and HNCACB datasets of human ubiquitin are provided (SECTION 2.2: "Contents of the NMRanalyst Distribution") for practising the 3D analysis of a larger protein. See http://www.sciencesoft.net/3DAnalysis.html for further instructions.

7.5 Incorporating NMRanalyst Programs in Other Software

NMR software typically includes a peak-picking program for the simplistic computerized spectral analysis. NMRanalyst's programs were written to be included in other software systems as "super-peak-picker". The NMRanalyst user interface is an example of how to integrate the analysis code into software packages. The names of the computing programs are: 1d_analysis, fft, analyze, report, assembleit, and graphic (the programs have a .exe postfix in the MS Windows version). The computing programs read input from the corresponding *.txt file (1d_analysis.txt, fft.txt, analyze.txt, report.txt, assembleit.txt and graphic.txt) and write output and error messages to the screen. The *.txt files are normal text files and contain one key-value pair each line. The key is the name of a variable to be set. Supported variable names are described in the corresponding workwindow reference chapter. The workwindow input screen provides a convenient way to edit the key-value pairs in a *.txt file, but any text editor can be used as well.

Click on the [UNIX Shell] button in the NMRanalyst window to start a UNIX shell window. Change the current directory to a desired analysis directory, such as the data directory used in CHAPTER 3: "Tutorial I: Using NMRanalyst". For spectrum type INADEQUATE, simply issue the command:

 % 1d_analysis INADEQUATE

The output from the analysis of the 1D spectrum, which normally appears in the workwindow output screen, is now written to the shell from which it was started. The computing processes of NMRanalyst can be run from any program or script which can issue operating system commands.

Please note that NMRanalyst programs are protected by the software contract. Even if incorporated in other software, they fall under the same distribution restrictions as the original NMRanalyst software.

7.6 The On-line Help System

The five tutorials contain the information to follow discussed examples. Further questions can be answered through the "NMRanalyst Manual" contained in the online help. CHAPTER 8: "Using the Help System" provides detailed information on how to use the help system. In the NMRanalyst window select [Manual...] from the [Help] pull-down menu, and the shown help window appears. The navigation section at the left side allows easy navigation among help topics. The [Help] menu at the top of the NMRanalyst window provides access to frequently used help topics.

1Courtesy of Paul A. Keifer, Ph.D., Varian Inc.



ScienceSoft, LLC
www.ScienceSoft.net
TOC PREV NEXT INDEX