Skip to the content.

Phase Correction in 3D Protein Spectra


In the 3D spectrum, phase correction is done for every dimension individually. SOLUTION 1 is only for the direct (F3) dimension, while SOLUTION 2 is for every dimension. I strongly recommend using them in combination.

Table of contents

  1. SOLUTION 1: Phase-Correct the 1st FID

  2. SOLUTION 2: Phase-Correct the F2-F3 projection of the 3D Spectrum

  3. Credits

Tested environment:


SOLUTION 1: Phase-Correct the 1st FID (F3)

The 1st (strongest) FID always corresponds to the direct (F3) dimension. Therefore, you cannot phase F2 and F3 with this solution.

Example Data: N-edited 3D NOESY of Carbonic Anhydrase 2 (29 kDa)

Download example data

  1. Reset all phase values to 0:

    Important: Look at the bottom of the pulse program to find the correct PHC0 and PHC1 values of F1-F3 - if there are any.

    • Set all phase values to zero with the following commands:
      3 PHC0 0
      3 PHC1 0
      2 PHC0 0
      2 PHC1 0
      1 PHC0 0
      1 PHC1 0
      
    • Reset spectrum truncation boundaries with the following commands:
      3 TDeff 0
      2 TDeff 0
      1 TDeff 0
      3 STSR 0
      2 STSR 0
      1 STSR 0
      3 STSI 0
      2 STSI 0
      1 STSI 0
      
    • You may want to increase the points in any dimension (SI) by x2 or x4 times (zero filling) to enhance resolution. Ensure that all SI values are powers of 2 (e.g., 1024, 2048, etc.).

Note: Linear Prediction is beneficial when only a few points have been recorded in a dimension, otherwise it may be detrimental and thus must be turned off.

  1. Extract the 1st FID:
    • Open the raw 3D spectrum and run rser eao 1 to extract the 1st FID followed by ft to switch to the frequency domain. Note that for other spectra, the eao argument may not be appropriate (see when to use rser 1 and when rser eao 1).
    • Alternatively, you can create a macro using the command edmac qfp.
    • In the editor that opens, write the following commands to save and process the spectrum:
       qsin
       fp
      
    • Run the command qfp.
    • In the pop-up window, set 1 at FID and 2 at PROCNO.
    • Issue twice .gr to display denser grid lines.

3D 1.FID

  1. Phase Correction:
    • Try first the automated 0th order phase correction with apk0 that will try to create an entirely absorptive lineshape in the spectrum.
    • If you are not entirely satisfied with the automatic phasing, enter .ph for manual phase correction. Set the pivot line (right mouse click) near the right limit to modify only PHC0, not PHC1.

      Note: Adjusting PHC1 (1st order correction) has minimal effect on frequencies near the pivot and no effect on PHC0.

    • Shift baseline to bottom with .sd and adjust its position with the arrow icons.
    • Adjust only PHC0 (0th order correction) so the FID line extends straightly and the highest peaks point upwards.

      Note: Avoid using automatic phasing (apk and its variants) as it generally doesn’t work well.

after phasing

  1. Save and Apply PHC0:
    • Note down the optimal PHC0 value (e.g., 151.40 degrees). Click the Save-disk nD icon to copy the values to the 3D spectrum, then click the Return icon. Double-click the processed 3D spectrum and ensure all PHC0 and PHC1 values except for F3(PHC0) are still set to 0.
    • Note: sometimes when the data was moved around, the Save-disk 3D button will complain on non-existing paths and wwwill not work. In those cases, one has to transfer the phasing coefficients manually.

correct phase parameters

  1. Initiate NUS-reconstruction:
    • Execute FnTYPE. If this option is not set to non-uniform sampling then you can safely use ftnd 0 nusft or just ft3d instead of ftnd 0 to skip the NUS-reconstruction in all relevant steps from this point on. But even if you run ftnd 0, Topspin will recognize it by itself and skip the otherwise tedious NUS-reconstruction.
    • Select the cs method within the NUS section of PROCPARS window or simply execute 3 Mdd_mod cs.
    • Execute the command ftnd 0 for Fourier transformation and NUS reconstruction - it shouldn’t last long in the 3D spectrum. This step helps in visualizing the spectrum and deciding on the dimensions to truncate.
  2. Check Visually the NUS-reconstructed Spectrum:
    • Click on the 23 icon and navigate the F2-F3 plane by holding left-mouse button on the double arrow icon and dragging your mouse. You will notice that there are no antiphase peak components in the F3 dimension.


SOLUTION 2: Phase-Correct the F2-F3 sum projection of the 3D Spectrum.

Summary of Commands and Steps

  1. Reset all phase values to 0:
    • Like step 1 in SOLUTION 1.
  2. Initiate Fourier Transform with zero phase values:
    • Like step 5 in SOLUTION 2.
  3. Create the Sum Projection of F2-F3:
    • Issue the following command to create the sum projection of F2-F3 (N-HN):
      sumpl 23 all all 23230
      

23 Sum Projection

13 Sum Projection

  1. Visualization and Dimension Truncation:
    • Open the 23 sum projection. The vertical line in the center represent water signal; the pulse sequences are designed to be symmetrical ensuring that the space before and after the water line is the same.
    • Zoom in by selecting the region of interest (RIO).

ROI 23 sum projection

1 ABSF1 1000
1 ABSF2 -1000
1 BC_mod no
2 BC_mod qpol
2 SSB 2.2

ROI 23 sum projection After Baseline Correction

  1. Prepare for Manual Phase Correction in the F2 dimension:
    • Enter .ph for manual phase correction.
    • Look for peaks that have an anti-phase component in the F2 dimension.
    • Place the cursor between the centers of these two peaks. Ensure the cursor is along the line connecting their centers.
    • Right-click and select “Add”.

Selected Misphased Peak

Selected Misphased Peak

  1. Enter Phase Correction Mode:
    • Click on the icon with the letter “R” to enter phase correction mode for the rows, which correspond to the F2 (HN) dimension in the 23 sum projection.
  2. Set the Pivot Point:
    • Right-click at the left limit of the screen and select “Set Pivot”.
    • Place the pivot point away from the peaks to avoid hiding the frequencies we want to phase.
    • Note: Only the PHC0 value will be adjusted, not PHC1, so the pivot point will not be used.

3D F2 Before Phasing

  1. Adjust the Phase:
    • Left-button press on the “0” icon and drag your mouse until the strongest signals appear at the top part of each panel.
    • Ensure the frequency lines from end to end look straight and balanced.

3D F2 After Phasing

  1. Save the PHC0 Value:
    • Note down the optimal PHC0 value (e.g., 151.40 degrees) and click on the “Save-disk” icon to save the selected PHC0 value to the 23 sum projection spectrum.
    • Click the Save-disk nD icon to copy the PHC0 value to the 3D spectrum.
    • Then click the Return icon.
  2. Verify the Phase-Corrected Spectrum:
    • The phase-corrected 23 sum projection should now appear without any anti-phase peak components.

3D Phased 23 sum projection

  1. Save and Apply PHC0 to 3D:
    • Switch to the 3D spectrum and run ft3d for FT and phase correction with the values you provided.
  2. Review and Adjust Peaks in 3D View:
    • Return to the 3D view to review the changes by clicking the cube icon.
    • Click on the /8 icon to display the peaks. Adjust the contour levels with your mouse.
    • Click on the aim icon to switch to 2D plane view and then to 23. Navigate through different planes using the up and down arrows icon. Identify a plane where peak phases are unsatisfactory. In this example, I couldn’t find a plane containing misphased peaks, but you got my point.

3D Problematic 23 Plane

  1. Extract and Process a Specific Plane:
    • Execute slice to extract the problematic plane.
    • A window will ask for the plane number (displayed at the bottom left in red letters; marked with green circle). Enter a Destination PROCNO, e.g. 23265.

3D Problematic 23 Slice

  1. Reconstruct Imaginary Dimension for Further Phase Correction:
    • Use xht1 followed by xht2 or navigate through the menu Advanced -> Special Transforms -> Hilbert in F1 and then Hilbert in F2 to reconstruct the imaginary dimensions.

Note: Files like 3rrr represent the spectrum in real space, whereas files such as 3iri, 3irr, 3rir, 3iir, and 3iii contain at least one imaginary dimension. Correspondingly, the command slice created a file 2rr in the 23265 folder. The Hilbert transformation created the files with imaginary parts: 2ri, 2ir, and 2ii.

3D Problematic 13 Plane

Phased 3D Spectrum


Baseline Correction

  1. Increase FID Dimensions:
    • Increase FID dimensions by x2, maximum x3 (e.g., if the dimension size is 60).
  2. Set Baseline:
    • 3 BC_mod spol
    • Choose a baseline where there are no peaks. In the proton dimension, set ABSF1 -> 9.0 ppm and ABSF2 -> -4.0 ppm.
  3. Fourier Transform:
    • Perform Fourier transform with ft3d.
  4. Contour Level Adjustment:
    • Click on the hill icon, set contour level sign to positive, Level increment to 1.2, and the number of levels to 30.
  5. Navigate and Adjust Peaks:
    • Navigate in depth (F2 dimension) by clicking the + icon until you reach two overlapping peaks.
    • Note down the plane number (e.g., 27/256).
    • Iterate with different FID sizes, window functions, and linear predictions until the peaks separate. Do not apply linear prediction on the proton dimension because many points are acquired, but it can be done in the indirect dimension. Avoid overdoing it.
  6. Baseline Correction:
    • Perform baseline correction with the commands:
        tabs1
        tabs3
      
    • This corrects the baseline in F1 and F3 dimensions.

Authors