Menu
▾
▴
#218 RFD: How Z-thickness should be calculated and displayed
**Opinion: Expressing the Z thickness in planes is fundamentally wrong and limiting
**
Assign 3.0.b2 expresses the z thickness of spectrum display modules in points or planes. In my opinion this is a fundamental flaw for windows that display spectra with a frequency or chemical shift axis in the z-dimension, although it has its place for differently sampled data such as pseudo nD data containing discrete (n-1)D subspectra that are resonably not to be treated as continuous, such as (perhaps arguably) relaxation time series, heteronuclear NOE with and without saturation, IPAP, etc. Peaks in spectra with continuous axes should not have to be placed on the the grid of data points, but can be at interpolated positions. To take an extreme example - if a point is exactly half way between two planes, which of the two planes should be displayed when the view is centred on that peak in z?
Consider viewing two 3D triple resonance experiments that have been recorded and/or processed such that they have different digital resolutions:
The optimal behaviour would be that one set of contours would be drawn for each spectrum describing the intensity visible in the 3D volume being viewed. Unfortunately it is probably too time consuming to continually recompute the summed intensities from multiple planes and redraw the contours.
The next best behaviour would be that the contours for planes of each spectrum visible within the defined 3D volume are displayed (the ANSIG and Analysis v1 & v2 way). This ensures that data from directly comparable volumes of the superimposed spectra are always displayed. One can set the minimum z-thickness to just above the digital resolution of the least well resolved spectrum to ensure contours for all displayed spectra are always drawn if that is considered desirable. Depth cueing of the planes (and peak markers) can be easily used in any 3D graphics language to improve useability.
The next best behaviour would be to always display a single plane - the one from each spectrum closest in z to the centre of the viewed volume. In this case the data displayed should be at least as closely comparable as possible, but anyone who has previously used software that can display an arbitrary thickness z-slice will be able to tell you the limitations this brings.
Displaying the z-thickness as a number of planes for any number over 1 results in an illogical situation where data from different volumes of the two spectra may be being compared with one another. If the digital resolution of the two spectra differed by a factor of a not at all unlikely 2, a z-thickness in planes that corresponded to say 2 ppm in the higher resolution spectrum would cover 4 ppm in the lower easily leading to confusion for the user.
RFD = request for discussion :)