home contents changes options help

last edited 15 years ago by damberger

Manual Structure Calculation using CARA.

This is a brief tutorial on how to use CARA together with structure calculation programs to manually calculate and refine structures. The manual approach is now supported by a number of user-written scripts. At the bottom of the page is a Quick summary of manual structure calculation using CARA.

NOTE: If you are doing a standard structure determination of a protein, then it is much more efficient (and more objective) to use the free (for academia) automated NOESY assignment program ATNOS/CANDID together with a structure calculation program like CYANA. In this case you would follow the instructions on the CARA wiki page for automated structure calculation.

Labeling of Protein

It is assumed that most sequential assignments have been completed and that the protein is 13C/15N-labeled but the instructions can be extrapolated to unlabeled or 15N-labeled proteins. I.e. with an unlabeled protein, one would open 2D NOESY spectra with HomoScope where the tools like "Propose Spin" are also available. However, in this case the spins forming the NOE (spinlink) must be assigned by the user in BOTH dimensions.

SpinLinks represent NOEs

The most important information for obtaining structural constraints in solution NMR are NOEs which indicate the close proximity of two atoms in space. NOEs are represented in CARA by SpinLinks. A spinlink is simply an entry with two SpinIds indicating which two spins are close in space. Any NOESY spectra where these two atoms can be correlated will display a crosspeak once the spinlink is created. I.e. CARA extrapolates the existence of the NOE in all relevant NOESY spectra. E.g. If you picked a spinlink for the NOE "GLU23 HA-ASN24 HN", then this NOE will appear in the GLU23 HA/CA strip of the 3D 13C-aliphatic NOESY and the ASN24 HN/N strip of the 3D 15N NOESY.

Inferred Peaks are not NOEs

By default, CARA displays all 1H spins belonging to a given spin-system in the NOESY tower strip. This is to help the user identify intraresidual NOEs. It is also helpful in matching up sequential peaks with the corresponding intraresidual peak. However, not ALL possible intraresidual NOEs are seen in the NOESY tower! E.g. the HZ of PHE may not be seen in the HN/N strip of the same PHE. To toggle on/off the display of these "inferred peaks", use "Strips-Show Inferred Peaks" in PolyScope or "View-Show Inferred Peaks" in StripScope. It is important to turn off the inferred peaks in strips when exporting a strip peaklist for structure calculation (otherwise nonexistent intraresidual peaks will be included in the list).

Assign NOEs using "Propose Spin"

To assign NOEs, you click on the unassigned signal in the NOESY strip (of PolyScope or StripScope) and then right-click "Propose Spin". A list of potential assignments will appear sorted according to how closely the candidate spins shift matches the current cursor position in the strip. Click on the most likely candidate.

E.g. In StripScope: In the ASN24 HN/N strip of the 3D 15N NOESY, you see that a signal lines up with the intraresidual HA-HN NOE of the GLU23 in the previous HN/N strip, and conclude that its probably the sequential GLU HA-ASN HN NOE. You click on the NOE and use "Propose Spin" to select the GLU23 HA from the list. The crosspeak appears.

You can check the consistency of this assignment by looking at the GLU23 HA/CA strip in the 3D 13C NOESY. CARA will display a crosspeak at the expected position of the NOE to ASN HN. If there is no intensity there, you may have to try another assignment. Checking for transposed peaks is facilitated by a script ProposeAndShow.lua which shows the relevant strips from NOESY spectra where the NOEs are expected.

If a reference structure is available, you can use the script ShowNoesInMolmol.lua together with Molmol to display potential NOE assignments on the structure.

Erase NOEs using "Delete Links"

You can delete to Spinlink you just created using undo or "zz". Generally you can erase a spinlink by clicking on the crosspeak and then right-clicking "Delete Link". Note that this is NOT the same as "Delete Spins" which deletes the spin that is linked to along with the link!

Create Peaklist and integrate it

Once you have assigned all the obvious NOEs, you can export the peaklist for structure calculation. Ensure that you have

This avoids writing out inferred (intraresidual) peaks that are not observed in the spectrum. However, it also means that you will have to use "Propose Spin" to create SpinLinks for all the intraresidue NOEs! These NOEs typically are used for calibration of NOE intensity to distances.

Export the Peaklist with "File-Export-Strip Peaklist to MonoScope". This will export the peaklist to MonoScope. You can inspect the position of the peaks and their identity to make sure nothing went wrong. Then setup the integration. See the wiki page on Integrator for details.

Export Peaklist for structure calculation

You can write out the integrated peaklist from MonoScope with "Peaks-Export Peaklist".

Export distance constraints for structure calculation (completely manual approach)

Alternatively, you can export distance constraints derived from SpinLinks or from PeakLists? to an external file. These will directly furnish constraints for structure calculations and would be used for a completely manual approach (instead of CANDID for automated assignment of NOESY peaklists).

CYANA structure calculations:

Use the script ExportSpinLinksToUpls.lua to convert SpinLinks into upper distance limits (UPLs) that are written to an external file. The script generates one UPL for each spinlink with a fixed distance of 5.5 Angstroms. This is very conservative assumption but may allow a global fold to be identified. A more sophisticated approach would include routines to calibrate the integrals vs. distances for each NOESY spectrum and generate a UPL file which used these calibrations. However. this would require a more complex script written by a motivated user. Note that a python script has been donated at the Tracker issue 0828 (not supported by CDT)

XPLOR-NIH structure calculations:

Spinlinks can also be exported to an XPLOR constraint file using the script ExportToXplor.lua. Again the simplifying assumption is made that all spinlinks correspond to a distance constraint of 5.5 A. There is also a script which generates XPLOR constraint files from PeakLists: PeakListToXplorConstraints.lua. Peaklists for the NOESY spectra should be present in the project. The script divides the peaks into four catagories based on intensities, calibrates intensities to distance, and writes out a constraint list.

Export Assignments and Sequence for structure calculation

You will also need to write out the assignments and the sequence for structure calculation. You can do this using the script "WriteAssignments.lua" available from the CALUA page of the wiki.

Set up structure calculation

You can now setup the script which executes the structure calculation with CYANA (or other program). It will need the location of the peaklists, the chemical shift assignments and the sequence.

Some Special Topics

Including unassigned peaks in the peaklist

Sometimes it is not possible to assign a peak unambiguously, but one would like to include it in the peaklist so that CANDID can assign it once structural information is available. The dilemma is that you can only create a spinlink to existing spins.

If you want to create a spinlink, you always need two spins since a spinlink is simply an entry linking two spins. Therefore you will have to create an unassigned spin and then link to it.

In PolyScope:

  1. You are viewing a strip of a system.
  2. In the strip: click on the NOE signal to the unassigned spin.
  3. Right-click in the strip and select "Pick Spin". This creates the unassigned spin IN the current spin-system. Note that when you pick the spin "Strips-Show Unlabeled Peaks" is automatically turned on so that the new spin is visible. If you open a new Scope, this will not be the case, and you will NOT see the spin. You will need to turn on "Show Unlabeled Peaks" which is in the Strips or View menu depending on which Scope you opened.
  4. Click on the newly created spin. Note the SpinId number from the status-command line: e.g. 
    Selected 1 peaks: 312:H/1843 29:I29.
    The ID number is "1843"
  5. Right-click and select "Propose Spin". Select the spin which has the ID number of the new spin. This creates a spinlink from the anchor spin of the strip to the new spin. Turn off "Strips-Show inferred" and you will still see the spin because it is linked to the anchor spin.

Note 1: You can skip step 4 in most cases. The cursor is exactly on the spin, so the top spin in the list is usually the spin you just created.

Note 2: Later when you find the correct assignment of the NOE, it is probably most efficient to delete the spinlink and the spin. Then use "Propose Spin" to correctly assign the NOE. If the correct spin was unassigned and you found the correct assignment with the NOESY in question, you can assign the spin itself. Use "sl" to show the SystemList?". Expand the system where the spin is located. Right-click on the spin and select "Assign to" and enter the the SystemId? of the System where you assigned the spin. Expand that system. Click on the spin and right-click "label spin" and give it the correct label.

Quick summary

  1. Complete the backbone and sidechain resonance assignments
  2. Use PolyScope to define SpinLinks by right-clicking on NOEs with the command "Propose spin"
  3. Turn off "Show inferred" and turn on "Show SpinLinks" to see only peaks due to SpinLinks
  4. Check for transposed peaks using the script ProposeAndShow.lua
  5. Generate distance constraints:

    a. From SpinLinks (all 5.5A constraints) using

    b. From PeakLists (calibrated constraints, currently only for XPLOR-NIH)

    1. Create PeakLists from NOESY spectra (export from PolyScope with "show inferred" OFF and "show spinlinks" ON)
    2. Integrate them with Integrator
    3. Convert peaks to distance constraints using PeakListToXplorConstraints.lua

Return to StructureCalculation

Return to FrontPage

subtopics: