Peak Filtering

Peak Filtering
Prev	Chapter 7. Peaks	Next

The NMRView peak picker, is fast but relatively simple, and this can lead to the inclusion of peaks that correspond to noise or artifacts. Instead of eliminating these during the initial identification of peak positions, NMRView provides a variety of routines for filtering out these peaks. Peaks can be deleted interactively via the GUI, both directly on the spectrum, and using the Peak Inspector or Peak Tables as described above. There also a variety of scripted procedures that are available for deleting peaks that are thinner, wider or less intense than specified criteria or are on the diagonal. Furthermore, simple Tcl scripts can be written by users to filter out peaks based on other criteria.

Here is a simple Tcl procedure (included in the standard NMRViewJ distribution) that deletes any peaks in a specified list whose intensity is below a specified threshold. The user could invoke this script by typing in the NMRView console ?Nv_PeakCutLow noesy 0.4?.

proc Nv_PeakCutLow {plist threshold} { set n [nv_peak n $plist] set ncut 0 foreachpeak iPeak $plist { set lvl [eval nv_peak elem int $plist.$iPeak] if {[ expr {abs($lvl)<$threshold} ] } { eval nv_peak delete $plist.$iPeak incr ncut } } puts "Cut $ncut of $n peaks" }

Two commands in the script are worth emphasizing. The foreachpeak command loops over all the peaks in a specified list (setting the iPeak variable to the peak number of each in turn). By using this command the user doesn?t need to worry about the number of peaks in the list or whether there are gaps in the peak numbering. The nv_peak elem command provides read and write access to all the parameters that characterize a particular peak. In this example, the nv_peak command is used to obtain the peak intensity, but all other elements of each peak such as chemical shift, width, volume, assignment label, and comments are accessible as well. Together these two commands can form the basis for scripts that perform a wide variety of operations on peak lists.

A graphical interface to some of the builtin filtering commands is available via the Peaks - Edit - Filter menu item. Documentation coming.

Peaks: Deletion, Compression and Gaps, or when is a peak really gone?

There are a variety of ways to delete peaks in NMRViewJ including: on the spectrum with the deletion cursor or delete selected peaks commands, from the Peak Inspector or Tables with the delete button, or with the low level nv_peak delete command. All of these actions don't actually permanently eliminate the peak. Instead they all set a status flag on the peak indicating that it should be considered as deleted. All peaks have a status value that is normally set to "0", but can be set to any integer value. Values less than 0 indicate a deleleted peak. Peak rescuscitation can be achieved by changing the status flag back to 0 (either by toggling the peak delete buttons or using the nv_peak undelete command.

But often you really want to be rid of the peaks forever. Perhaps you've established that they're just noise and you don't want them around cluttering up your analysis. Peak list compression will do the trick by searching through the peak list and permanently removing any peaks marked for deletion (status flag less than 0).

Every peak in NMRView has an identification number. When a peak list is first created the peaks are numbered sequentially starting at zero. When we delete peaks and then compress the list the peaks retain their original peak numbering. That is if we have three peaks, #0, #1 and #2, and then delete peak #1, we now have two peaks, #0, and #2. There is now a gap in the sequential numbering. There is absolutely no harm in this, but you may wish to update the numbering so that a gap-free sequential numbering again obtains. Now you'll have two peaks, #0, and #1, the new peak #1 is the original #2 peak.

Peak compressing and degapping is available from the Peak - Edit menu

Prev	Up	Next
The Peak Table	Home	Systematic Peak Examination