Index:
The present chapter will describe how you use the cross peak locating tools of the Pronto program. In general, users at the Carlsberg Laboratory prefer to do the peak picking of 2D spectra manually. In some cases, however, using the 2D cross peak identifier built into Pronto will give good results, but in all cases you still have to go through the whole spectrum and check for missing peaks and for false peaks, and that process could in some cases take longer time than to locate the cross peaks manually. In the 3D case, the combination of using the automatic cross peak locator and manual removal of peaks in noise areas is normally quicker than locating all the peaks manually, as going through a whole 3D spectrum takes a long time.
The manual cross peak picking has been described previously. The principle is, that you go through the spectrum, for instance by plotting 1.05x1.05 ppm areas around (11.5,11.5), (11.5,10.5), (11.5,9.5), etc. The reason for using an area larger than 1x1 ppm is to make it possible to locate peaks in the border area of the contouring diagram. Select the Add Peak function in the contour plot window. Click on the center of each of the peaks in the window to add the peaks to the cross peak catalog. If markers do not appear in the contouring window, you probably forgot to select the display of markers for that spectrum in the contour zoom window.
The currently supported cross peak identifier utilizes symmetry recognition (J. C. Hoch, H. Shen, M. Kjær, S. Ludvigsen, and F. M. Poulsen, Carlsberg Res. Commun. 52, 111pp (1987)). The algorithm consists of three steps: Locating points of interest, calculation of the projection operator, and locating maxima in the projected map.
To find a proper set of parameters for locating peaks, start with a small region. First contour the region to analyze.
Open the 2D cross peak identifier from the tools menu. A window as shown in Figure 1 is shown. Click in the spectrum to analyze in the field at the top. Some of the parameters to fill in depend upon what type of spectrum you are analyzing, and some are general. The low and high parameters at the bottom are used to specify the range of the spectrum to analyze. After the spectrum has been filled in at the top, the axes names will change to what you have specified in the spectrum zoom window. Fill in the lower and higher chemical shift values for the two axes. Select a small region to begin with, and contour out the same area so you can follow the progress of the cross peak identification. The peak markings, however, will not appear on the contour diagram until the whole search is finished, but if the search was not successful, press the Regret last search button and start again with some new parameters. Note: Pressing this button erases all the peaks made by the last search, independent of if this search was made for another spectrum on another day. It will never erase peaks inserted manually.
Pressing the Go button starts the search using the current parameters. A window with a dial pops up, and you can follow the progress of the search. The search can be aborted, if it finds unreasonably many peaks in the area being analyzed. When you press Continue in the dial window after the search, the peaks will be inserted into the database, and if the proper region is displayed in a contouring window, shown inside that. However, this requires that you have selected marking of cross peaks in the zoom spectrum for the contouring setup.
The peak search contains three phases, as described below.
The filter type is selected with the Point-Of-Interest type parameter, and currently three types are supported: ALL: all points are considered interesting, MAX: all points above the cutoff, Cutoff 1, are interesting, and BITMAP: Points inside a +/-/-/+ pattern, where the numerical values of the peak components must be larger than the parameter Cutoff 1. An area of a width and height of Range 1 Hz is considered for the search of this pattern. The BITMAP method only works for COSY-type spectra.
The projection operator will be applied to all points that are within the distance of Scan Hz from points that are considered interesting. The projection operator is selected using the Projection operator parameter. Three projection operators are currently supported: D2 (for COSY type spectra), C2v (for NOESY type spectra), and D2d (special case: Works with COSY spectra where the peaks have the same separation constants in the two dimensions). The way the projection operator works, depends upon the Projection type parameter:
MEDIAN The projection is calculated by taking the sum of the average of the two middle values of the absolute values of the symmetrically related points, i.e. if the four symmetrically points have the values: -5.9, 7.2, 6.9, and -11.0, the value 0.5*(6.9+7.2) is added to the sum.
MEAN The means of the absolute value of the four symmetrically related points are summed.
MINABS The minimum of the absolute value of the four points is summed.
MAX The maximum value of the absolute value of the four points is summed.
BOLTON This is more special, look in the Hoch article, formulas 5 and 6! This works best, but is slower.
The size of the rectangle over which the projection operator is applied, is set with the Grid parameter, in Hz.
The values in the projected map that correspond to positions outside the points of interest are set to zero.
The peaks themselves are determined as maxima of the values calculated by the projection operator: For a position to be considered a peak, the value of the projection operator must be greater than Cutoff 2, and the values within Range 2 Hz must be less than the center point. If the parameter Calculate Separation Constant is set to YES, the values for the separation constants of the peak are calculated from the location of the maxima of the projected points. If one of the separation constants is greater than the value of Maximum Separation Constant (in Hz), the peak is deleted.
Use the default Point-Of-Interest type, Projection operator, and Projection type parameters to start with. Select Cutoff 1 to be above the noise level of the spectrum. This can be done by looking at the contouring levels used to display a region without too much noise. If you use the Normalize facility in the spectrum zoom window, the noise level should be around 1.0. The value of Cutoff 2 is more difficult to predict, as it depends upon the projection operator and type, and the size of the region in which the projection operator works, but in general, the value should be larger than Cutoff 1. To get a feeling of the order of the values, make a search with a value of Cutoff 2 that is the same as the value of Cutoff 1 and look at the range of the peak intensities showed in the cross peak zoom window. Select some of the false peaks (if any) using the mouse in the contouring window, and look at the value of the Intensity field. This is directly related to the Cutoff 2 value. If the program cannot locate a particular peak, even though you try with very low values of the cutoffs, try running the program with another (or no) Point-of-Interest type, however this could lead to many false peaks. If you set the Projection type to Bolton, the cross peak identifier will reject peak candidates, if the four subpeaks do not have nearly the same intensity, with different signs. This could remove some false peaks in noisy areas, but will take longer time.
Selecting calculation of separation constants gives a rough value for a separation constant in the two dimensions. For a more accurate value of the coupling constant and separation constant, use the Coupling Constant Measuring Tool instead.
The projection operator assumes that you have an anti-phase pattern where the upper right and the lower left subpeaks are positive. If this is not the case, either rephase the spectrum or select the Inverse field to Inverse instead of Normal.
To avoid locating peaks near the diagonal, specify a non-zero value in the Exclude diagonal field. Peaks where the differences between the two ppm values are less than the value in this field will be excluded.
The Point-of-Interest type must either be set to ALL or MAX. The Projection operator should be set to: C2v. Do not calculate the separation constant, as this will be nonsense.
The algorithm used for locating 3D cross peaks is much simpler than the one used in the 2D case: It simply looks for a local maximum above a certain threshold.
The setup window for the 3D cross peak identifier is shown in Figure 2. Click in the spectrum to be analyzed in the field at the top. The Cutoff value is the minimum value of the center of a peak. If you specify a negative value, the cross peak identifier will look for both positive and negative peaks, which have an absolute value above the absolute value of the cutoff. Set a cutoff value above the noise level. You can use the normalize facility in the spectrum zoom window to determine the noise level, in combination with contouring of selected areas of the spectrum. The ranges specified below is the number of points away from the center point in which the values must be below the center point. Normally you will use the value of one in all three dimensions. When a spectrum is inserted in the field at the top of the window, the lowest and highest ppm values in the ranges are set to the minimum and maximum value available in the spectrum.
Pressing the Go button starts the search. Currently, there is no way to stop a 3D search, so start with a small area of the spectrum to find the right parameters to be used. If you must abort the cross peak search, it can be done by pressing control-c in the window where Pronto was started: This will kill the whole program and leave you with a ruined database, so saving the database before starting the cross peak search is a good idea. Currently, a regret last button does not exist for the 3D cross peak identifier, but you can use the one available in the 2D cross peak identifier.