While it is possible to analyze NMR data by directly decomposing the raw data into lists of parameters such as frequencies and linewidths, most users still rely on Fourier transformation of the FID and then identifying peak positions in the transformed data. Accordingly, NMRView provides many tools for locating and analyzing these spectral peaks. The PeakPick Panel is the standard starting point for generating lists of peaks in a spectrum.
![]() |
Once you've picked your peaks, you'll notice a few things. First, you'll find the name of your peaklist in the peak list identifier window, between the List pulldown and Pick button. This is the active list for picking purposes, and can now be selected in the List... pulldown. Second, you'll see a bunch of rectangles with crosshairs will be drawn over your peaks in your spectrum display. Also, you'll see that the number of peaks picked becomes displayed in the Peak Peak panel. If you're appending to an existing list, this number will be the number of new peaks, not the total number of peaks in the list.
To pick peaks in a particular region of the spectrum, display the region in any spectral window. Then choose PeakPick from that window's pop-up menu. The PeakPick Control Window will appear. Type the name that you wish to be associated with the new peaklist in the text field of the control window. Click on the Pick button to start the automatic peakpicking. As peaks are identified in the spectrum, boxes annotated with the peak number will be drawn on the spectrum at the position of each identified peak.
Some spectra, especially 3D and 4D spectra, may be collected with a sweep width narrower than the frequency range of the spectral peaks. In these spectra, peaks are folded to a position that differs from their true position by an integer multiple of the sweep width. Peaks on the edge of such spectra "wrap" around to the opposite edge. At present NMRViewJ wraps for peak searching at all times. Thus, the peakpicker will wrap at the edge of the spectrum so that the peak width and position are properly determined.
To pick peaks in a specified region of the spectral window, place the crosshair cursors so as to box in the desired region. Select the Box attribute in the PeakPick Control Window. Now, only the region within the box will be analyzed for peaks. To pick peaks in the entire window select the Window attribute instead. The peaks that are picked may be stored as a new peak list, they may replace an existing peak list, or they may be appended to an existing peak list. The desired option may be selected from the pulldown list that specifies the New, Replace, or Append Attributes.
Not yet supported in NMRViewJ. The filter option under pick region restricts the peak pick to regions of the dataset defined by a previously acquired peak list. For example one might peak pick an hncacb spectrum looking in regions of the spectra that correspond to the 1H and 15N coordinates of an HNCO spectrum. If the list that is used for filtering contains close or overlapping peaks this will result in the same reqion of the active dataset being peak picked multiple times. To prevent this first remove the overlapping peaks in the filter list using the "couple" function under the edit menu in the peak analysis tool.
Enter the name of the peaklist. If mode (see below) is "New" there must not be an existing peaklist with this name. If mode is "Append" or "Replace", there must be an existing peaklist with this name.
Specify whether to pick the entire region of the displayed in spectral window, or just that within the crosshair box.
Peakpick the entire region displayed in spectral window.
Peakpick the entire region within the crosshair box
FIXME Not yet implemented in NvJ. Peaks will be picked in regions defined by a pre-defined peak list. Selecting Filter causes the Filter List portions of the window to open. Once a peak list has been selected the Filter Axes and Correlate Axes Input sections open.
Specify the peak list that will be used to control the filtered peak pick.
Axes of the active dataset that is to be peak picked using the filter.
Specify which axes of the filter peak pick list are to be used to control the filtered peak pick.
Specify whether to create a new list, replace an existing one or append to an existing one.
Picked peaks will go into a new peaklist.
Picked peaks will replace an existing peaklist.
Picked peaks will be appended to the end of an existing peaklist
Value is zero for a 2D peak pick. Set to non-zero value for 3D-peak pick when displaying only a 2D plane of the 3D dataset. Set to zero to peak pick a series of 2D planes in a 3D dataset (e.g. when relaxation data is collected as 2D planes combined in a 3D dataset).
Do the peakpick.
Indicates how many peaks were picked.
NMR datasets often have regions that have much higher noise levels than other areas. These are often areas with systematic artifacts, for example a stripe from residual water. A threshold that is appropriate for most regions of the spectrum may be completely inapprpriate in these artifact rich areas and if they are included in the peak picking process a large number of peaks will be found that do not represent actual resonances from the targeted molecule. NMRViewJ now includes a method to determine a local estimate of the noise level at all points in the spectrum. The peak picker can be set to only identify points that are both above the general threshold for the whole spectrum, and above a specified multiple of the locally calculated noise.
The local noise estimate is calculated by estimating the noise in vectors that intersect from all dimensions at each data point. The noise estimate from the dimension that has the largest noise estimate is used as an estimate of the local noise at that peak. Calculating these estimates is quite fast for 2D spectra, but can be somewhat time-consuming for higher-dimensional spectra.
Local noise thresholding is turned on by setting the Noise value in the PeakPick tab to some value greater than zero. A value of 10.0 seems to be a reasonable choise. After setting this value click the Pick button. If a local noise estimatiion has not been done for the spectrum you will be prompted as to whether to do the calculation. At this point estimates are not saved between NMRViewJ sessions. You can pick multiple times in one session, trying different values of the Noise multiplier each time, without needing to regernate the noise estimate.
Here is a region of an HSQC spectrum where the peaks have been picked without Local Noise Thresholding. Note the streak of peaks at the position of the residual water.

Here is the same region where the peaks have been picked with a noise multipler of 10.0 Note that, even though water region was included in the peak picking, no peaks were found there. Not shown here, but users will find that "real" peaks that are quite close to a streak will be found.
