Felix User Guide

Preference pulldown

Preference/Plot Parameters

This menu item sets various display parameters that affect the plotting of spectrum data.

Note

Various items appear on this menu, depending on whether the data is 1D, ND, or molecular.

1D plot parameters

For 1D data the parameters are divided among five control panels: Basic, Axis, Tick, Place, and Stack, which can be activated from each other by clicking the appropriate button.

Table 6. 1D plot parameters--Basic

Control	Function
Stack Depth	Determine the number of data buffers plotted. A value of zero means to plot the work vector alone; a value of 1 means to plot the work vector and the first data buffer.
Stack Overlap	How much the plotted vectors overlap: a value of 0 means no overlap between the work vector and data buffer vectors when plotted; a value of 1 means to arrange the plots so they overlap completely.
Stack Order	Choose whether the displayed data buffers are arranged from the top down with buffer one at the top, or from the bottom up with buffer one near the bottom, immediately above the work vector plot. The work vector is always at the bottom of the plot.
Color Scheme	Select a color from a list. You can select define if you want some other color.
Color Number	Set the initial color for drawing. The default pen colors are shown in Table  7.
Color Cycle	Define the number of colors used to plot the 1D work vector and data buffer vectors. The first vector is drawn in the color specified by Color Number. If Color Cycle is set to 0 or 1, all the vectors are drawn in the same color. When Pen Cycle is set to 2, the second vector is drawn with a pen color identified by the index Color Number +1, the third vector is drawn with the color Color Number, and so on.
Plot Mode	Whether real values, imaginary values, or both are plotted.
Plot Type	Whether the data values are plotted as points or lines.
Center Plot	Setting Center Plot to on sets the center of the y-axis to zero.
Absolute Intensity	Setting Absolute Intensity to Yes locks in the plot scaling factor that Felix calculated for the previous plot. That plot scaling factor ensured that the plot filled the vertical display space. In effect, Absolute Intensity turns off automatic scaling, which allows you to compare the magnitudes of two different data sets or the magnitudes of the data in the work vector and the data buffer vectors.
Scale Factor	Set a multiplier that scales the data plots. The default setting is 1, which causes Felix to best-fill the vertical space of the plot. A setting of 0.5 produces a plot of half the size.

ND plot parameters

Similar to 1D data, plot parameters for ND data are divided among five control panels: Basic, Axis, Tick, Place and Stack.

The Basic control panel contains tools that control the appearance of the plotted matrix values. You can for example, choose the base contour level to plot, the number of contour levels to plot, the level multiplier, and the colors of the contour levels.

Table 10. ND plot parameters--Basic

Control	Function
Contour Threshold	Set the base (lowest) contour level. Base contour level value is the product of Contour Threshold and the reserved symbol mscale (see the Felix Command Language Reference Guide).
Automatic	Automatically find the threshold.
Number of Levels	Specify the number of contour levels to plot.
Level Multiplier	Determine the difference between values of consecutive contour levels when Number of Levels has a value greater than 1. When Interval Mode is Geometric, a contour level is the product of the previous contour level and the Level Multiplier (the Level Multiplier must be greater than 1). When Interval Mode is Linear, a contour level is the sum of the previous contour level and the Level Multiplier (the Level Multiplier must be greater than 0).
Negative Levels	Whether contours are drawn for data values less than zero. If you want negative contour levels drawn, set Negative Levels to On; otherwise, leave it set to Off. This parameter is most often set to On for contour plots of COSY matrices.
Color Scheme	Specify a predefined color scheme: Fire Ramp: 28-color ramp for positive-only data. Blue/Green: 16-color ramp for positive and negative contours. Green Ramp: 16-color ramp for positive-only contours. Blue Ramp: 16-color ramp for positive-only contours. Red Ramp: 16-color ramp for positive-only contours.   For any of these predefined color schemes, the Color Number, Color Cycle, Number of Levels, and Negative Levels are inactivated. For the next 6 choices (Red, Green, Blue, Magenta, Cyan, Yellow) you can set the Number of Levels and Negative Levels. For the last choice (Define) you must manually set all 4 variables (Color Number, Color Cycle, Number of Levels, and Negative Levels).
Color Number and Color Cycle	Pen index for the first contour level. To display more than one contour level, the colors for subsequent levels cycle through the next Color Cycle pen indices. When Color Cycle is 0, all contours are the same color. For example, with Number of Levels set to 4, Color Number set to 1, and Color Cycle set to 2, the first contour level has color index one, the second has color index two, the third has color index one, and the fourth has color index two. The default pen color indices are shown in Table  7.
Interval Mode	Whether adjacent contour levels are increased by a multiplicative factor, Geometric, or an additive term, Linear.
Interpolation Mode	Method by which Felix calculates the position of the contour between matrix data points. None is the default and draws faster than Spline. Set Interpolation to None unless you are expanding a very small plot region and do not like the boxy look of the contour plot vectors. The Spline mode yields smoother looking contours.

The ND Plot Parameters--Axis control panel contains tools that control the appearance of the other portions of the plot--the axis units for each dimension, projection display, scale factors for each dimension, annotations, box around the plot, and grid lines.

Table 11. ND plot parameters--Axis

Control	Function
Scale D1 and Scale D2 and Scale D3 and Scale D4	Set scaling factors used to modify the proportions of the plot. When the scale factors are both 1, Felix proportions the plot to the number of points in each dimension of the displayed region. For example, if one dimension has twice as many points as the other, the plot is rectangular with one side twice the length of the other. To display the plot as a square, set the second scale factor to 2 (or set Scale D1 to 0.5). When the scale factors are set to -1, then Felix automatically scales each axis to fill the frame.
Axis D1 and Axis D2 and Axis D3 and Axis D4	Axis units for each dimension of the matrix: None, Points, Hertz, and PPM. Before displaying the matrix with Hertz or PPM axis units, you must reference the matrix.
Draw Annotations	Set this parameter to Off if you want no annotations on the plot and to On if you want annotations drawn each time you make an intensity or contour plot. To produce annotations on a hardcopy plot, this parameter must be On.
Annotation File	Name of the file used to draw annotations. You must create this file using the Edit/Annotation menu item.
Row Projection	Whether a 1D plot appears along the x axis and the contents of the plot. The possible settings for Row Projection are Off, Sum, Shadow, Buffer 1, Buffer 2, and Buffer 3. To turn off the 1D plot, select Off. To display a projection calculated as the sum of the values along each column of the matrix, select Sum. To display a projection calculated as the maximum data value along each column, select Shadow. To display data stored in one of the first three buffers, select Buffer 1, Buffer 2, or Buffer 3.
Column Projection	Whether a 1D plot appears along the y axis and the contents of the plot. The possible settings for Column Projection are Off, Sum, Shadow, Buffer 1, Buffer 2, and Buffer 3. To turn off the 1D plot, select Off. To display a projection calculated as the sum of the values along each row of the matrix, select Sum. To display a projection calculated as the maximum data value along each row, select Shadow. To display data stored in one of the first three buffers, select Buffer 1, Buffer 2, or Buffer 3.
Proj Size (fract)	Set the size of the 1D projection plots. Specify Proj Size as a fraction of the full display size. For example, if you want the matrix to occupy 90% of the display and the 1D projection to occupy 10%, set Proj Size to 0.1.
Default Plot Type	Set the default plot type, which can be Intensity, Contour, Stack, or Null. The Plot Type menu item also sets the Default Plot Type.
Draw Levels on 1D	Whether current contour levels are drawn on 1D plots, which assists with setting contour levels. Set Draw Levels on 1D to On and load a row or column of the matrix into the workspace using the Display/1D Vector Mode menu item. The current contour levels appear as horizontal lines on the 1D plot.
Box Around Plot	whether the plot is enclosed within a box. By default, Box Around Plot is set to Yes, and Felix draws a rectangle around the plot.
Grid Spacing	An integer that defines the number of grid lines per label to draw on the plot. By default, Felix draws no grid lines, so Grid Spacing is set to 0. When a positive integer is used, the grids are solid lines through the plot. When a negative value is used, tick marks are placed on the outside edges of the plot.

The ND Plot Parameters--Tick control panel allows you to control the tick mark positions and axis text size for each dimension in a multi-dimensional data set.

Table 12. ND plot parameters--Tick

Control	Function
Tickmark Type	Whether tickmarks are displayed in the Default   manner or are User adjustable.
Major Tickmarks	Adjust the separation between major tickmarks for each axis.
Minor Tickmarks	Adjust the separation between minor tickmarks for each axis.
Axis Text Scale	Adjust the relative size of Axis Text.
Pen Width	Adjust the pen width of Axis Text.
Text Slant	Adjust the slant of Axis Text.
Text Thickness	Adjust the text thickness of Axis Text.

The ND Plot Parameters--Place control panel allows you to position the plot in the current Felix frame.

Table 13. ND plot parameters--Place

Control	Function
Manual Plot Size	Set Manual Plot Size to on to define the size of the plot by specifying values for the X Size and Y Size settings. When Manual Plot Size is set to No, Felix draws the plot to fill the space between the X Offset and Y Offset and the upper-right corner of the display area.
X Offset	The left edge of the plot relative to the Felix frame, in inches.
Y Offset	The bottom edge of the plot relative to the Felix frame, in inches.
X Size	The width of the plot in inches.
Y Size	The height of the plot in inches.

The ND plot parameters--Stack control panel allows you to change the settings for the stack plot.

Table 14. ND plot parameters--Stack

Control	Function
X Axis Skew (fraction)	Skew applied to subsequent stack plots as a fraction of the x-axis size. To improve plotting speed, the skew value is converted to an integral number of points. A positive value skews the plot to the right and a negative value skews to the left.
Y Axis Skew (fraction)	The distance, as a fraction of the y-axis size, between the first and last row of a stack plot. To avoid clipping large peaks near the back of the plot, set to a value less than one.
Row Increment	Stack plot increment value. For example, if Row Increment is 2, every second row in the 2D matrix is plotted. Larger values plot faster and show more space between rows but may cause small peaks to be missed. Smaller values plot more slowly and often leave too space between rows.
Height Scaling Multiplier	Vertical scale factor. By default, the factor is 1, so Felix draws the stack plot to fill the vertical space of the display. The Height Scaling Multiplier adjusts the vertical size of the stack plot.
Maximum Peak Height	Maximum peak height, in inches, for a peak drawn in a stack plot. Peaks above the specified height are clipped so that you can see behind them.

Preference/Levels

As an alternative to control panel adjustment of ND plotting parameters, a real-time interface that contains buttons and sliders enables you to adjust the display. This interface is accessed by selecting the Preference/Levels menu item (<Alt>-ps). You can adjust the Contour Threshold, the Number of Levels, and the Level Multiplier in real time. To exit select the Quit button.

Preference/1D Scale

As an alternative to control panel adjustment of 1D plotting parameters, a real-time interface that contains buttons and sliders enables you to manipulate parameters. You can activate this feature by selecting the Preference/1D Scale menu item (<Alt>-pe).

You may adjust the spectrum's vertical scale by dragging the Scale slider or entering a value to the box next to it. You can toggle between absolute and relative scaling. You can adjust the vertical offset (Offset) or, if there are multiple 1D slices to display, you can adjust the vertical overlap between them (Overlap).

You can specify any point's (peak's) vertical position by clicking the Set Scale button and clicking the cursor somewhere in the spectrum display. The point at that position will have the vertical value of the cursor.

To exit the real-time interface, select the Exit button.

Preference/1D Slice

With the Preference/1D Slice menu item (<Alt>-p1) you can set what direction(s) the 1D vectors should take from the current matrix in the current frame and to which frame(s) those vectors should be exported. Turning the Arrange Frames toggle on means to allow Felix to arrange the newly created frames. If you decide not to let Felix arrange the frames, you may do it yourself later.

After this preference is set, you can use the View/Draw 1D Slices, View/Multiple 1D Slices, or View/Thick 1D Slice menu items to export 1D vectors from the current frame. The setup is valid only for the current matrix in the current frame.

Preference/2D Slice

With the Preference/2D Slice menu item (<Alt>-p2) you can set the direction(s) the 2D planes should be take from in the current 3D or 4D matrix and to which frame(s) the orthogonal planes should be exported. Turning the Arrange Frames toggle on means to allow Felix to arrange the newly created frames. If you decide not to let Felix arrange the frames, you may do it manually.

Once this preference is set, you can use the View/Draw 2D Slices menu item to export 2D planes from the current frame. The setup is valid only for the current matrix in the current frame.

Preference/Overlay

The Preference/Overlay (<Alt>-po) menu item sets the experiments from which to plot the overlay contour plots on each other. Overlay plots are particularly useful for examining several related spectra (for example, DQF COSY or TOCSY). Preference/Overlay is used to define which spectra from the project to overlay, so you can use it only in conjunction with Assign.

Preference/Reference

To reference a spectrum, select the Preference/Reference menu item (<Alt>-pr). Different control panels appear, depending on the dimensionality of the data.

1D spectrum referencing

The Reference 1D Data control panel opens to prompt you for the referencing information. The Spectral Frequency and Spectral Width must be set to the values of the spectrometer frequency in MHz and the spectral width in Hz. You can enter the Reference Point value either by typing it into the box or by clicking the Cursor button in the control panel, moving the vertical cursor to the desired reference point on the plot, and clicking the left mouse button. The Reference 1D Data control panel reappears and you enter the reference value in the appropriate entry box, e.g., if you want axis units of Hertz, enter a reference value in the Reference Hertz box. For PPM units, enter a reference value in the Reference PPM box. Finally, select OK to close the panel and redisplay the data with the selected axis units.

ND spectrum referencing

The Reference Matrix control panel provides tools to assist you with referencing your matrix. The Axis type may be set to PPM, None, Points, or Hertz. Next enter the Spectral Frequency in MHz, the Spectral Width in Hz, the Reference Point in data points, and the Reference Shift in PPM, then select OK. You may select the Reference Point for a displayed dimension with a mouse click by using any of the three Cursor buttons.

Preference/Pick Parameters

Two different control panels appear for setting peak-picking parameters when you select this menu item, depending on the dimensionality of the data.

1D Spectrum

Before picking peaks, set the threshold value manually or via the cursor. Selecting the Cursor option in the popup and selecting OK gives you a line cursor for selecting the peak-picking threshold on the plot. In this control panel you can also specify if you want only positive peaks, only negative peaks, or both (Selection Mode). You can also specify the peak table and the units that should be displayed on the peaks (Points, ppm, Hz, or the Assignment). One-dimensional peak markers can be drawn as lines, arrows, or line and arrows (Draw Style).

ND Spectrum

The control panel provides access to the cross-peak entity (table), the cross-peak selection mode (positive peaks, antiphase peaks, negative peaks, and positive and negative peaks), and tools for specifying the antiphase search window size, fixed footprint sizes, and minimum and maximum halfwidth values. The Info button provides help in setting the picking parameters. Antiphase Peaks mode searches for antiphase multiplet peaks, as in a COSY matrix.You can select the regular peak picker (which is faster, but sometimes less reliable) or the Stella peak picker (which is slower, depends on example peaks, but, with a good training set, can give superior peak set). The method for interpolating an extremum should be set (by fitting a second-order polynomial (Quadratic Interpolation) or Center of Gravity).

Stella Peak Picker parameters

The general rule is that more example peaks means longer execution time: for a 2D matrix it can take up to several minutes to pick peaks using 6-10 example peaks. For 3D or 4D sets the time can be even longer. A local maximum is picked as a peak if:

• Its maximum exceeds the current threshold.
• It has at least a certain number of neighbors whose intensity exceeds the threshold.
• It matches one of the example peaks with a match factor (-1 or +1) that exceeds the Minimum match factor threshold (if example peaks are used).

The definable parameters are:

Table 15. Stella Peak Picker parameters

Control	Function
Selection Mode	Pick positive, negative, or both types of peaks. If you pick negative peaks then you must have negative example peaks, and if you pick positive peaks you must have positive example peaks.
Local maximum method	Set the local maximum-defining method. You can use the rough maximum, the center of the gravity, or interpolation.
Minimum match factor	Should be set between -1 and +1, to distinguish between genuine peaks and spurious peaks based on matching the provided example peaks.
Hump tolerance factor	Used in peak bounds detection.
Extra points for peakbox	Choose certain points to be added to the found peak limits.
Neighbors above threshold	Define at least how many neighbors must exceed the current threshold for the local maximum to be considered a peak.
D1 search, D2 search, ...	Select how many neighbor points should be checked for local maxima in each dimension (D1 search, D2 search, ...). These search parameters are used to define a box in which the center should be local maximum to be considered as a possible peak. The neighbors are also considered within this box (for a 2D matrix, if D1 search = 2 and D2 search = 2, then a square containing 25 points is considered at one time).
Limits	Whether you want to pick the full spectrum or just a subset. If you set Limits to Define, then after selecting OK a new control panel shows up where you can define the limits of the region you want to pick, in points or ppm.
Output Level	The amount of information to print at each possible peak (Quiet, Low, Medium, or High).

Preference/Integral

This menu item (<Alt>-pl) sets parameters that affect the integration of 1D spectra and the display of the integrals.

Preference/Peak Display

The Preference/Peak Display (<Alt-pd>) menu item controls the appearance of the ND cross peak footprints by setting the display switch, the cross peak entity, and other relevant parameters:

Table 16. Preference/Peak Display parameters

Control	Function
Draw Crosspeaks	Off if you want no cross peaks on the plot and On if you want cross peaks drawn each time you make an intensity or contour plot. To draw cross peaks on a hardcopy plot, this parameter must be on.
Crosspeak Entity	The name of the entity used to draw cross peaks. You must create this entity with menu items within the Peaks pulldown.
Crosspeak Symbol	The marker to be drawn at each cross peak: a cross (its size is Half Width Factor X 2 X crosspeak_width) with a surrounding box (Default), just a cross (Cross Only), a half size cross (Small Cross), just a small box (Small Box), or just the number (Number).
Label Peaks	Whether cross-peak footprints are labelled with nothing, the peak item number, the assignment names, or (for proteins) a shorthand notation.
Label Size	Size of the cross-peak labels in inches.
Half Width Factor	Size of the cross-peak footprint, together with the half-widths of a peak. The width of the cross-peak footprint displayed is twice the product of the half-width and the Half Width Factor. The footprint size influences the results of many Felix menu items that operate on cross peaks.
Coloring Mode	Cross-peak footprint coloring mode: one color or (for 3D and 4D spectra) based on the peak position compared to the current plane, or (for assigned peaks) based on assignment or on whether they belong to a specific prototype pattern. The latter two options are available only with Assign.
Color	Cross-peak footprint color. Specify the color as a pen index, using the Define option and entering the number in the Color Number box (see Pen Number) or referring to it by name.

Preference/DQF Parameters

The Preference/DQF Parameters (<Alt>-pa) menu item lets you choose an optimization method for the J-coupling measurement in DQF-COSY spectra (quasi-Newton, Simplex, or simulated annealing).

Preference/Frequency Display

The Preference/Frequency Display (<Alt>-pq) menu item lets you specify the horizontal or vertical line colors for frequency drawing.

Note

Preference/Keypad

The Preference/Keypad (<Alt>-pk) menu item lets you specify the small step size in points for <Alt>-keypad navigation.

Preference/Frame Layout

The Preference/Frame Layout menu item (<Alt>-pf) contains frame layouts for creating a new set of frames (New Layout options) or for rearranging the existing frames (Rearrange Layout options).

Note

The New Layout options delete the contents of existing graphic frames. To avoid this use the Rearrange Layout options.

Table 18. Rearrange Layout Options

Control	Description
Cascade	Arrange the existing frames in a cascade manner.
Tile	Arrange the existing frames in a tile fashion.
Horizontal	Resize the existing frames to maximize the horizontal dimension and arrange them vertically.
Vertical	Resize the existing frames to maximize the vertical dimension and arrange them horizontally.

Preference/Frame Connection

With the Preference/Frame Connection (<Alt>-pc) menu item you can connect and disconnect up to 12 frames containing different views of different spectra. You can use this connection, for example, to view a ¹⁵N-¹H HSQC spectrum in the primary frame and a ¹H-¹H view from the corresponding ¹⁵N-separated TOCSY and NOESY spectra in the two secondary frames, having the two secondary frames connected along each dimension and defining the plane selection direction in the HSQC spectrum to be along the ¹⁵N dimension. Therefore, looking at each H_N-N peak in the HSQC spectrum, you can bring up the corresponding ¹⁵N plane in the TOCSY and NOESY spectra and visually collect each spin system quickly by hand.

You can also connect two frames, each containing a strip along an H_N frequency from an HNCACB and a CBCACONH spectrum at the same ¹⁵N frequency. This allows you to quickly scan through the entire two 3D spectra and find the intra- and inter-residual CB-CA-N-HN.

You can select a trivial connection between two frames (D1 to D1 or D1-D2 to D1-D2) or specify a more elaborate one using the Custom option.

After a frame connection is set, you can temporarily disconnect them by using the Disable option. Then you can restore the connection with the Enable option.

Preference/Multiple Cursor

This menu item is similar to the Preference/Frame Connection menu item, in that you can connect and disconnect the cursors using the Preference/Multiple Cursor menu item (<Alt>-pu). You must specify which axis of one frame should share cursor positions with other frame's axis.

Preference/Table

With the Preference/Table menu item (<Alt>-pt) you can specify the current table (entity) names for peaks, integrals, volumes, and other items.

Preference/Directory

The Preference/Directory menu item (<Alt>-py) opens a control panel showing the current directory prefixes for each type of file that Felix uses.

You can edit any box to change a file prefix. The file types that Felix uses are listed in Table  19. The notation "read only" implies a sharable directory, while "read+write" denotes a directory that should be owned and used by only one person.

Table 19.Felix directories

Control	Type	Description
Data	read+write	1D data files.
Matrices	read+write	2D and ND matrix files.
Database	read+write	Felix database files.
FELIX Macros	read only	Felix macros.
FELIX Dialogboxes	read only	Felix control panels.
FELIX IF Macros	read only	Felix user interface menu files.
FELIX Icons	read only	Felix user interface icon files.
User Macros	read +write	User-written macros.
NMRRefine DB	read+write	Insight II files.
Limits	read+write	Plot limits files.
Annotations	read+write	Annotation macro files.
Parameters	read+write	Program context saved parameter files.
Messages	read only	Felix error message files.
DB Schema	read only	Database schema prototype files.
Coordinates	read+write	Atomic coordinate files.
Text Files	read+write	All written ASCII text files.
Runtime Files	read+write	Various temporary runtime files.
Foreign Data	read +write	Spectrometer data files to be filtered.
Filter Images	read only	Data filter programs spawned by Felix.`

Preference/Memory

The Preference/Memory menu item (<Alt>-pm) opens a control panel showing the currently defined buffer size and the number of buffers allocated for Felix. You can change the memory allocation.

Process pulldown

Process/DC Offset

A problem you can encounter when zero filling time-domain data is related to baseline correction. If the right-most data points in the FID are significantly different from zero, and zero filling is performed, a discontinuity (step function) is introduced into the data. The spectrum resulting from the Fourier transformation of data that contain a step function has wiggles or waves in its baseline.

It is also especially important to avoid discontinuities when zero filling badly truncated data. This common problem is encountered when processing multi-dimensional data, though it can be prevented by using baseline correction to remove DC offset.

Manipulation of time-domain data prior to Fourier transformation can be used to change the size and appearance of transformed spectra. In addition, some type of spectrum artifacts can be eliminated. Raw FID's are usually corrected before Fourier transformation to remove any DC offset that may have occurred during data acquisition. This is done by setting a baseline correction fraction with the control panel that appears when you select the Process/DC Offset (<Alt>-pd) menu item. The baseline-correction fraction specifies the fraction of the FID, starting from the right side, to be averaged to eliminate the DC offset. The default value of this symbol is 0.2, based on the assumption that most of the signal has decayed to zero in the last 20% of the FID. By averaging the last quarter or so of points in the FID, a good zero level can usually be defined. For complex time-domain data, which contain both real and imaginary parts, the DC offset for each part is calculated independently.

If the baseline-correction function can calculate an accurate zero level, the effect on the transformed spectrum will be to eliminate a spike at the observed frequency. However, if the data are badly truncated (not enough data points were collected), baseline correction may not be able to calculate an adequate zero level. In fact, by applying baseline correction you may add a DC offset. If you are worried that your data may be truncated but still want or need to baseline correct the DC offset in your FID, try baseline correcting using a smaller fraction of the FID; that is. set the value of the baseline correction-fraction to 0.05.

Process/Zero Fill

The Process/ZeroFill menu item (<Alt>-pz) allows you to zero-fill spectra. Zero filling is commonly performed on time-domain data. By selecting the Process/ZeroFill / BC menu item, you may increase the number of points in the transformed spectrum and thereby increase the spectrum's apparent digital resolution. Zero filling a spectrum defaults to doubling the size of the data, but you may zero fill to any desired size.

Process/Solvent Suppression

Solvent signal suppression

NMR data are frequently composed of signals arising not only from resonances of interest, but also from the solvent used to dissolve the sample. Solvent signals may compromise the analysis of the signals of interest, and in extreme cases may completely obscure important spectrum features.

Although effective methods for minimizing the intensity of solvent signals at acquisition time exist, e.g., through tailored excitation, post-acquisition methods can be extremely useful when undesirable solvent signals persist.

Felix offers three methods for reducing the intensity of such solvent signals: a linear prediction-based algorithm (LP), a convolution-based method (CNV), and a polynomial-based method.

LP-Based solvent suppression

The linear prediction-based solvent-reduction routine exploits a technical feature of the LP algorithm to estimate and remove contributions from the most intense components in the spectrum (the intensities of the signals present in the interferogram are effectively ranked). The algorithm relies on the fact that solvent resonance frequently represents the most intense component in the spectrum (as with data acquired in H₂O) and explicitly assumes this as a part of its function. If the signal identified as the most intense component is not significantly larger (by default, 5 times the value of the other components), no solvent-peak elimination is done.

Linear prediction-based solvent reduction is accessed in Felix by selecting the Linear Prediction option for Method in the Process/Solvent Suppression menu item. A control panel appears, asking you for the number of data points to use in the LP calculation and the number of signals to remove. A value of 1 for Signals To Remove eliminates only the most intense component of the spectrum, a value of 2 removes the two most intense components, and so on. If the signal identified as the most intense component is not significantly larger than the other components, no solvent-peak elimination occurs. You can view the results of solvent suppression and change the parameters interactively if you specify Real-Time for the Method.

CNV-based solvent suppression

The convolution-based solvent-reduction routine conducts a convolution of the data with a sinebell or Gaussian function to first identify the lowest-frequency component, and then subtracts that component from the data (Marion and Bax 1989). Two parameters are available in this control panel: the convolution function, which can be either a sinebell or Gaussian function (which at best is largely an empirical issue) and the function width. The best value for the function width depends upon the widths of resonances in the spectrum and the resolution. In practice its value is empirically derived.

Convolution-based solvent reduction is accessed in Felix by selecting the Time-Domain Convolution option for the Method parameter in the Process/Solvent Suppression menu item. A control panel asks you for the convolution function type (sinebell or Gaussian), and the function width (the default value of 10 works well for ¹H data acquired in 1-2 K data points). You can view the results of solvent suppression and change the parameters interactively if you specify Real-Time for the Method.

Polynomial-based solvent suppression

The polynomial-based solvent-suppression method uses a polynomial fitting method to remove solvent signals from the time-domain data. The solvent signal is approximated by calculating the mean value of groups of data points and fitting a polynomial to these mean values. The resulting function is then subtracted from the time-domain data. This technique works best when the solvent frequency is close to zero.

Polynomial-based solvent suppression is accessed in Felix by selecting the Polynomial option for the Method parameter in the Process/Solvent Suppression menu item. A control panel asks you to enter the Points to Use and the Polynomial Order. The Points to Use represents the number of data points in each group of points to average. The Polynomial Order represents the order of the polynomial that is used to fit the set of average points. You can view the results of solvent suppression and change the parameters interactively if you specify Real-Time for the Method.

Process/Window Function

Time-domain NMR data can be multiplied by window functions that perform digital filtering to reduce noise or increase spectrum resolution. For example, the noise level in 1D NMR data can be attenuated by multiplying the FID by an exponential window function.

The Process/Window Function menu item (<Alt>-pw) allows you to select a window function and adjust its parameters by entering parameters directly. You can also set the function interactively while Felix displays plots of both the window function and the product of the FID (possibly the FT'd spectrum) and the window function.

Real-time adjustment

The available window functions are: Sinebell, Sinebell^2, Skewed Sinebell, Skewed Sinebell^2, Exponential, Gaussian, Trapezoid, Kaiser, and Matched. Once you select a window function, Felix opens a new control panel where you can enter the parameters and apply the selected window function or can select the real-time option. If you select the real-time option, a real-time interface panel appears, which consists of six buttons and a group of sliders, depending on the window function. Using the buttons, you can Expand the displayed FID or display the full FID (Full). You can apply FT or draw only the FID using the popup (No FFT/FFT/Bruker FT/Digital FT). You can also Reset the parameters to their original values. When you finish viewing the window function, you can leave the menu item by selecting Keep or Quit. These buttons close the real-time interface and either retain the FID as it appeared with the window function or restore the original FID without the window function applied, respectively.

The sliders provide a way to directly adjust the window function parameters in real time. For example, the real-time interface for the Sinebell window function has sliders for the Window Size and the Phase Shift. You can adjust these parameters by moving the cursor over the slider and dragging. The red slider bar moves and the updated value is displayed within the slider.

The plot display is controlled through the keyboard: you can increase or decrease the scale of the plot or move the axis left/right and up/down, as well as zoom in and out.

The Felix display presents an on-the-fly updated plot of the window function and the product of the data values and window function as you adjust the window parameters with the real-time panel.

Window function descriptions

Matched filter

Matched filter is an automatic version of exponential multiplication that examines the FID and chooses an appropriate Lorentzian broadening. The matched filter calculates and applies a matched exponential window to the FID. The line broadening is calculated by performing a least-squares fit to the FID. If the FID has an extremely low signal-to-noise ratio, the fit may fail, and a message to that effect appears on the screen. Note that one large, narrow, softened resonance may dominate the fit. After applying the matched filter, the global line-broadening parameter is set to the value of the line broadening that was applied. The matched filter menu item is useful because it allows Felix to determine the optimal line-broadening parameter for your spectrum, and thus gives the best signal to noise ratios.

To access this function, select it from the interface or enter the following at the command line:

>	mf

For more detailed information, please see the mf command in the Felix Command Language Reference Guide.

Convolution difference

Convolution difference is an apodization function that calculates the difference between no line broadening and specified line broadening.

To access this function, enter at the command line:

>	cd lbroad




lbroad  

Adjusts the convolution parameter for the exponential.  

For more detailed information, please refer to the cd command in the Felix Command Language Reference Guide.

Sinebell

To access this function, either select it from the interface or enter at the command line:

>	sb wsize wshift




wsize  

Adjusts the number of data points for the window function.  


wshift  

Adjusts the phase shift of the window function.  

Sinebell squared

To access this function, either select it from the interface or enter at the command line:

>	ss wsize wshift




wsize  

Adjusts the number of data points for the window function.  


wshift  

Adjusts the phase shift of the window function.  

Skewed sinebell

To access this function, either select it from the interface or enter at the command line:

>	qsb wsize wshift wskew




wsize  

Adjusts the number of data points for the window function.  


wshift  

Adjusts the phase shift of the window function.  


wskew  

Adjusts the skew of the window function.  

Skewed sinebell squared

To access this function, either select it from the interface or enter at the command line:

>	qss wsize wshift wskew




wsize  

Adjusts the number of data points for the window function.  


wshift  

Adjusts the phase shift of the window function.  


wskew  

Adjusts the skew of the window function.  

Exponential linebroadening

To access this function, either select it from the interface or enter at the command line:

>	em lbroad




lbroad  

Adjusts the line-broadening parameter for the exponential.  

Gaussian linebroadening

To access this function, either select it from the interface or enter at the command line:

>	gm lbroad gbroad




lbroad  

Adjusts the line-broadening parameter for the exponential.  


gbroad  

Adjusts the Gaussian parameter for the exponential.  

Trapezoidal

To access this function, either select it from the interface or enter at the command line:

>	tm p1 p2 p3




p1  

Adjusts the first point of the trapezoid.  


p2  

Adjusts the second point of the trapezoid.  


p3  

Adjusts the third point of the trapezoid.  

Kaiser

To access this function, either select it from the interface or enter at the command line:

>	kw wsize alpha




wsize  

Adjusts the number of data points for the window function.  


alpha  

Adjusts the alpha parameter of the Kaiser window.  

Process/Linear prediction

Linear prediction estimates the value of a point based on the values of adjacent points. This can be used to replace corrupted values in an FID or to extend an FID.

First-point prediction

The Linear Predict First menu item uses linear prediction to replace data values at the beginning of the FID. The Points to use tool in the control panel defines the number of points used to calculate the LP coefficients. A reasonable value for this parameter would be the number of points in the workspace minus the number of predicted points (the larger the setting of Points to use, the longer the action takes to complete). A good value for the Number of coefficients setting is one quarter to one third the value of Points to use. The Number of peaks is included for compatibility with older macros, but is not used in the calculation. The Number of points to predict specifies the point at which the First Points function begins predicting values. It estimates data values from that point backward to the first point of the FID. For example, to replace the values of the first three points of the FID with predicted values, enter a value of 3 for First Points.

Last-point prediction

The Linear Predict Last menu item can be used for predicting first points, extending the FID, or replacing corrupted points. The First Point tool in the control panel defines the start of points used to calculate the LP coefficients. The Last Point parameter defines the end of points to be used to calculate the LP coefficients. The Start Point parameter defines the start of points to calculate. The End Point parameter defines the end of points to calculate. A good value for the Number of coefficients setting is one quarter to one third the value of points to be used for prediction. The four choices for the Method are Backward, Forward, Forward-Backward, and Mirror. You can use root reflection by turning the Use Root Reflection parameter on. The Type of mirror LP is used only when the Method is set to Mirror. The 90-180 method is used when the data collection is delayed by one half the dwell time. The 0-0 method is used when there is no delay in the data collection.

Process/Transform

The menu items under Process/Transform (<Alt>-st) apply to Fourier transformation, linear prediction, and Hilbert transforms in the workspace.

Complex FFT

The Complex FFT option applies a complex Fourier transform to the data in the work space. For this transform, the data must be true complex data, characterized by simultaneous sample and conversion of the real and imaginary signals.

Bruker FFT

The Bruker FFT option performs a complex Fourier transform on complex data that are unique to some Bruker spectrometers. These spectrometers cannot sample and convert the real and imaginary signals simultaneously; instead, they collect the real and imaginary signals alternately. If your data were collected in this mode, you must use the Bruker FFT option in the Process/Transform menu item.

Real FFT

The Real FFT option performs a real Fourier transform on real data in the work space. After the real Fourier transform, the data become complex with the spectrum in the real part of the work space.

Oversampled FFT

The Oversampled FFT option performs a complex Fourier transform on digitally oversampled data collected on Bruker DMX and newer-series spectrometers. If your data were collected using digital oversampling you should use this menu item to do the transform.

Inverse FFT

At times you may need to convert frequency-domain data into time-domain data. For this purpose, use the inverse Fourier transform by selecting the Inverse FFT option in the Process/Transform menu item.

Hilbert transform

You can perform a Hilbert integral transform on the data in the work space by selecting the Hilbert Transform option in the Process/Transform menu item. The Hilbert transform is valuable for creating a complex spectrum from a real spectrum, that is, it transforms real data in the frequency domain into complex data in the frequency domain. The Hilbert transform is required for rephasing a spectrum after the imaginary part is discarded, which can occur with multidimensional NMR data processing.

Process/Phase Correct

After Fourier transformation, a spectrum often appears to be out of phase--that is, the resonance lines appear to be a mixture of absorptive and dispersive shapes. This is due to several factors, including finite pulse lengths, acquisition delays, and analog filter response. NMR spectra can be phase-corrected after transformation by multiplying each data point value pair by a phase factor. This menu item can also be accessed with the hot keys <Alt>-sp.

Real-time phase correction

One of the most valuable features of Felix is real-time phasing capability. This feature can be activated by selecting the Real-Time option in the Process/Phase Correction menu item.

To demonstrate this feature's capabilities, follow these steps to generate an un-phased spectrum:

• Read the file sample.dat (select the File/Open menu item and select sample.dat as the Selection).
• Apply a matched filter window function (select the Process/Window Function menu item and select Matched Filter in the control panel).
• Perform a Bruker FFT (select the Bruker FFT option from the Process/Transform menu item).

The spectrum now displayed on the graphics screen is that of a hairpin DNA dodecamer. To activate the real-time phasing interface, select the Real-Time option from the Process/Phase Correction menu item. A set of sliders and buttons appears. By default, the entire spectral width is displayed (see Figure  2).

Figure 2 . Real-time phase correction interface

The real-time phase interface includes two active slider bars. To adjust the current values for the zero-order and first-order phase corrections, drag the respective slider bars. Releasing the mouse button stops adjusting the slider bar and updates the current parameter. The current values of the zero-order phase and the first-order phase are displayed to the right of their respective slider bars. To adjust the scale of your data, use the keypad.

To set the pivot for your spectrum, right-click at the pivot position while holding the <Shift> key pressed in either the small or the main window. The current position of the pivot is indicated by a small vertical red line at the bottom of your spectrum. If the pivot is outside the current display region, the pivot indicator appears at the edge of your spectrum.

For expanding the spectrum, you can use the left mouse button while holding the <Shift> key pressed. Since your data most likely contain more data points than there are pixels on the screen, it is not uncommon for peaks to disappear on the real-time phasing display. This is especially true for ¹³C spectra where the resonance lines are very narrow.

• To expand the current region, drag the small crosshair cursor to define the region of interest. This function is activated by left-clicking in the spectrum while holding down the <Shift> key or by clicking the Expand button.
• To display the full spectrum, select the Full button from the real-time phase interface.
• To change the sensitivity for phase 0 and phase 1 manipulation to a greater value, select the Coarse button. Each selection of the button increases the sensitivity of the horizontal mouse actions by 20% for those values.
• To change sensitivity for Phase0 and Phase1 manipulation to a less sensitive action, select the Fine button. Each selection of the button decreases the sensitivity of the horizontal mouse action by 20% for those values.
• To specifically set the range of the Phase0 and Phase1 slider bars, select the Parameters button.
• You can directly enter phase parameters in the boxes to the right of the sliders.
• To quit the real-time interface without saving or applying the phasing parameters, select Cancel.
• To quit the real-time interface and save and apply the phasing parameters, select Keep. This updates the reserved symbols phase0 and phase1.
  

  Figure 3 . Phase-Corrected spectrum

Phase correction using parameters

In addition to using the real-time phase interface, you may also phase a spectrum by manually setting values for phase0 and phase1. Select Parameter as the Method in the Process/Phase Correction menu item.

The values in this control panel are updated when you exit the real-time phasing interface. To phase a spectrum manually, you must first define parameters for the zero-order correction (phase0) and the first-order correction (phase1). Felix does not use a separate value for a spectrum pivot; instead, this is incorporated into the values of phase0 and phase1. To apply a phasing correction, update values for phase0 and phase1 in the control panel and simply select OK. To exit the control panel without applying or updating the phase parameters, select Cancel.

When you repeatedly process similar spectra and want to apply a known set of phase corrections to a spectrum, it is easier to enter the phase corrections in this control panel and apply them in this manner than it is to re-phase a spectrum interactively in the real-time phasing interface.

Automatic phase correction

In addition to the phasing techniques described above, Felix provides several functions for automatic phasing of a 1D spectrum. To use one of these methods, select the Automatic option in the Process/Phase Correction menu item, then select one of the four Auto Method options.

The automatic phasing methods include the PAMPAS and APSL methods for spectra with non-split peaks, such as decoupled ¹³C and DEPT spectra; a method based on peak integration, for general in-phase 1D spectra; and a basic method intended for common proton spectra.

Except for the basic method, you can specify one or more excluded areas, to exclude solvent peaks when calculating the phase parameters. For the PAMPAS and APSL methods, you can also specify a Filter Width, which is the minimum peak width (at the peak bottom) required for a sample peak to be used in the calculation of phase parameters.

Felix selects the default values for Auto Method and Filter Width automatically, based on the spectrum data in the workspace; however, you can change these settings.

Process/Baseline correction

The Process/Baseline Correction (<Alt>-sb) menu item contains options that deal with baseline points or do baseline correction.

Process/Baseline correction/Auto Pick Points

To define baseline points, select the Auto Pick Points option in the Process/Baseline Correction menu item. This automatically generates a list of baseline points. Display markers for each baseline point picked in the spectrum are shown at the bottom of the current spectrum.

Process/Baseline correction/Auto Pick Points w/FLATT

The Auto Pick Points w/FLATT option of the Process/Baseline Correction menu item uses the FLATT algorithm (Guntert 1992) for selecting baseline points in a spectrum. The resulting points are stored in the entity whose name is stored in the symbol basent. You are first asked for the Basepoint Line Width to use in calculating the chi value for the spectrum. You are then asked for the Baseline Width, Minimum Chi Square, Factor(tau), and Stride to use in selecting the baseline points. For a more complete description of the required parameters, please see the abp and chi commands in Appendix A of the Felix Command Language Reference Manual.

Process/Baseline correction/Pick Points via Cursor and Manual Pick Points

You may also add baseline points singly using the Pick Points via Cursor option or the Manual Pick Points option of the Process/Baseline Correction menu item. You can select each baseline point via a cursor or by manually typing in the desired points via a menu interface. In using the cursor, points are added by left-clicking the desired baseline points with the crosshair cursor. To exit this mode, click outside the spectrum.

Process/Baseline correction/Delete All Points

To delete all the baseline points, select Delete All Points in the Process/Baseline Correction menu item. This deletes the current baseline points entity from the database; it requires confirmation via a dialog box.

Process/Baseline correction/Delete Points in Region

If you make a mistake while selecting individual baseline points or if you want to modify the current list of baseline points, you may delete a region of points using the graphical interface. First, select Delete Points in Region in the Process/Baseline Correction menu item to create a small crosshair cursor. Then drag out a region of baseline points to delete.

Process/Baseline correction/Baseline correction

Once the baseline points are defined, you can choose one of several baseline-correction algorithms:

Process/Baseline correction/Polynomial

The baseline-correction algorithm generates smoother baseline correction functions from baseline points. The Polynomial correction option of the Process/Baseline Correction menu item differs from the cubic spline correction algorithm in that the baseline does not necessarily pass exactly through each baseline point, but a best fit is calculated. In addition, the order of the polynomial (from 2 to 9) can be set in the polynomial control panel. A polynomial of order two yields a smooth parabolic function, and a polynomial of order nine generates a more complex correction function. A polynomial of an order between three and five is usually sufficient to give accurate baseline correction.

Process/Baseline correction/Real-Time Polynomial

The Felix real-time baseline-correction feature lets you adjust the coefficients of a polynomial baseline function while displaying both the resulting baseline function and baseline-corrected spectrum superimposed. When you select the Real-Time Polynomial option of the Process/Baseline Correction menu item, Felix asks you to specify a polynomial order for the correction function and an interval width, which is used to average baseline point values as described earlier. Select OK, and the real-time baseline correction interface appears. When you finish correcting the baseline you can exit the interface and keep the corrected spectrum (Keep) or Cancel the interface and restore the original spectrum. The displayed region along the x axis of the spectrum can be altered by clicking Expand and using the small crosshair cursor to drag a box around the desired region. You can restore the complete spectrum by clicking Full. If you are dissatisfied with any of the baseline points, which are indicated by red ticks below the spectrum, you can add or remove points by clicking Add Points or Delete Points. When you are satisfied with the baseline points, click Fit to automatically calculate the polynomial coefficients (the calculated baseline appears as a red line superimposed on the spectrum) and click Apply to apply the correction to the spectrum. The polynomial coefficients can be adjusted individually in real time--after selecting the individual coefficient through the popup next to the slider--with the slider located along the bottom of the interface. Again, when the red baseline appears to coincide with the spectrum baseline, click Apply to correct the spectrum. You can zero the polynomial coefficients by clicking Zero and restore the original spectrum with the Reset button. In addition, the displayed spectrum can be shifted and stretched vertically with the keypad.

Process/Baseline correction/Cubic Spline

The cubic spline algorithm, applied by selecting the Cubic Spline option from the Process/Baseline Correction menu item, generates a baseline that passes exactly through each baseline point. A cubic spline may yield a kinked baseline if the defined baseline data points are close together and noisy. To minimize this problem, the interval width reserved symbol iwidth may be adjusted to a number larger than 1. Increasing the interval width minimizes the kinked baseline problem by averaging the data values in an interval of points around each picked point and using that average value as the baseline point.

Process/Baseline correction/Automatic w/ ABL

A baseline-correction function supported by Felix that does not require explicit baseline points is accessed by selecting the Automatic w/ABL option from the Process/Baseline Correction menu item. This automatically selects noise points and performs a baseline correction for each point. You need to input values for the noise level and the peak size in points. Depending on the number of points in your spectrum and your line widths, these values may need to be adjusted several times to fit your data. Therefore, it is recommended that you save a nonbaseline-corrected spectrum before applying the correction. This algorithm was reported by Dietrich et al. (1991) and implemented by W. Massefski.

Process/Baseline correction/Automatic w/ FLATT

Felix also provides the FLATT baseline-correction algorithm, a technique introduced by Guntert and Wuthrich (1992). The FLATT algorithm automatically finds baseline segments in the spectrum and uses linear least-squares to fit a truncated Fourier series to these points. To use FLATT, select the Automatic w/FLATT option in the Process/Baseline Correction menu item. Felix asks you for the Basepoint Line Width, which is used to calculated the minimum chi-square value. Enter an integer with a value that is small but larger than half the width of the widest peak. When you select OK, Felix determines a value for the minimum chi square, which should correspond to the contribution of noise to the chi-square value. This value is displayed in the next control panel, which opens automatically and asks you for baseline-correction parameters. You need to enter a value for the Baseline Width as you did for the minimum chi-square estimate. You may adjust the Minimum Chi Square value if you want. The control panel also asks you for the Points to Correct, which is the number of Fourier series terms used to fit the baseline, and the Factor (Tau), which specifies how much larger than the minimum chi-square value a segment's chi-square value can be and still be considered baseline. When the chi-square value of a segment exceeds the product of the minimum chi-square value and the Factor (Tau), the segment is considered to contain peak information and is rejected as baseline. Select OK to perform the action.

Process/Baseline correction/Automatic w/FaceLift

Felix also provides the FaceLift baseline-correction algorithm (Chylla & Markley 1993). This signal-recognition based utility automatically identifies baseline points and subtracts the baseline points from the original spectrum data.

To use FaceLift, select the Automatic w/FaceLift option in the Process/Baseline Correction menu item. Felix asks you for the Filter Width, which is the half-width of the smoothing data window over which datapoints are sampled. The half-width determines the minimum line width of artifacts that will be removed from the spectrum. The recommended range is 32-64 datapoints (powers of 2 are not necessary). The Number of Standard Deviations is used to determine a threshold standard deviation, above which any point is considered to be a signal point. The recommended range is 2.5-3.0. Clicking OK executes the FaceLift algorithm.

Process/1D Data Processing

The Process/1D Data Processing menu item (<Alt>-s1) allows you to interactively process 1D data or the first FID of an ND data set. The first control panel that appears allows you to specify the Filter Type for the kind of data you want to process. In general, if you are working with raw spectrometer data you should specify All Files (BRUKER, VARIAN, JEOL:*) as the Filter Type.You can then navigate through the desired directories to get to the data. At this point you want to select the actual spectrometer datafile. This is usually an FID or SER file. To select the FID you can double-click the filename or click the filename and then select OK.

Selecting OK opens another control panel containing the 1D header menu parameters, which are taken from the header information of the spectrometer datafile. Be sure to check that the proper parameter values are displayed and correct them if necessary. The Data Size parameter is in complex points if the Data Type parameter is Complex. Spectrometer data generally have a Data Type of Complex. The Bruker Data Type is used only  for Bruker QSEQ data where data points are collected alternately as on some older Bruker spectrometers. When you are satisfied that the header parameters are correct, select OK.

You are now presented with the main 1D data-processing control panel, whose controls are grouped into several sections. The top section lists the individual processing options. These options are similar to those used for 2D/3D/4D data processing. See Table  20 for more information on the individual processing options.

Table 20. Transform Parameters

Control	Description
Dimension to Process	Specify which dimension to process. For the 2D/ND menus you can choose standard processing or reprocessing. You might want to reprocess a dimension if you had first processed it without linear prediction. After fully processing the matrix, you could reprocess a dimension and do linear prediction by using an option such as D2HT+IFT+LP+FT. This does a Hilbert transform on the data in D2, inverse transforms it, linear predicts, and then does a normal ft. If you want to reprocess a dimension later on, you should not use a window function during the initial transformation of that dimension.   These macros generally assume the following correspondence between the Felix matrix dimensions D1/D2/D3 and the experimental acquisition dimensions:   For 2D Data For 3D Data For 4D Data D1 --> t2 D1 --> t3 D1 --> t4 D2 --> t1 D2 --> t2 D2 --> t3 D3 --> t1 D3 --> t2 D4 --> t1 The Felix D1 dimension always corresponds to the acquisition dimension.
Output Matrix Filename	Name of the matrix file that will contain the processed data. This file is built by the D1 transform.
Load Matrix in Memory	Load the entire matrix in memory instead of keeping it on disk. If you have enough memory this can significantly decrease the amount of time required for processing (not available for the D1 dimension).
Dimension 1 Size	Number of points in the D1 dimension of the matrix file (which is directly related to the digital resolution in D1). This number must be a power of 2.
Dimension 2/3/4 Size	Number of points in the indicated dimension of the matrix file, (which is directly related to the digital resolution in that dimension). Processing considerations mandate that these size parameters must be at least as large as the number of data records collected along that dimension. Thus, for a given number of complex points collected along one of the indirect dimensions the Dimension Size parameter must be at least twice as large as the number of complex points in order to hold all the data records collected in that dimension. This number must always be a power of 2.
Processing Mode	Specify how the data are processed. If you select bundle then the data are processed on the local workstation using the Felix "bundle" mode. If you want to process the data using several workstations then select distributed.
Correct DC-offset	Remove a DC value from the time-domain data, provided that the data points represented by the Fraction parameter are actually zero ± noise. The Fraction parameter indicates the decimal fraction of the final part of the data vector to use for the DC-offset calculation, that is, 0.2 indicates that the final 20% of the FID is used in the calculation that determines how much DC-offset to remove.
Correct 1st-point	Correct the initial data points using linear prediction or by a simple scaling of the first point (for D2 or D3 processing). If the initial dwell time in the virtual time domain of an ND experiment is something other than one-half the normal value, a correction to the value of the first point in the D2(t2) vector and/or the D3(t1) vector may be desirable (Otting et al. 1990). The Correct 1st-point parameter allows you to specify a simple scaling factor of one-half or to employ linear prediction to reconstruct the first data points.
Zero Fill	Specify the final data size for 1D processing (not applicable to 2D/3D/4D processing).
Solvent Suppression	Three types of solvent suppression are available: a linear-prediction-based algorithm, a convolution-based method, and a polynomial-based method.   Parameters for the convolution-based method, the LP-based methods, and the polynomial-based method are discussed in detail in Process/Solvent Suppression and in the Felix Command Language Reference Guide under the cnv, lps, and pso commands, respectively.
Window Function	Specify what type of apodization function to use. The options for Window Function are discussed in detail in Window function descriptions.
Linear prediction	Whether to augment the acquired data using a straightforward LP-based extension or the "mirror-image" trick (Zhu and Bax 1990), which is particularly appropriate for data acquired using "constant-time" evolution in t2 and/or t1. The goal is a slight improvement in resolution.
FT Type	The D1 transform menu item provides options for four types of Fourier transforms: Real (rft), Complex (ft) for true complex data, Bruker (bft) for Bruker data acquired using alternately sampled real and imaginary data points, and Oversampled (dft) for Bruker digitally filtered data. If Oversampled data is specified, then the appropriate Decim Factor and DSPIVS FACTOR are taken from the parameter files.
Phasing Mode	The selection for Phasing Mode depends upon whether or not appropriate zero (Phase0) and first-order (Phase1) phase parameters are known. If such values have previously been determined, then you may set the parameter value to Use Parameters and supply appropriate values for the phase parameter input boxes. If the phase parameter values are not known, the Interactive Mode opens the real-time phasing interface (see Real-time phase correction). If you select the Interactive mode, you can enter the number of the FID's you would like to phase in the FID to Phase box.
Baseline Correction	Please see Process/Baseline correction for a thorough discussion of the baseline correction utilities.
Absolute Intensity	Set absolute intensity mode on, which means that the spectrum scaling is set by the last displayed spectrum. This allows you to change various processing parameters and still observe the effect on the intensity of peaks.
Reverse vector	Certain hypercomplex phase-cycling protocols effectively render the complex-conjugate of what is normally expected by the ND States processing utility--such data appear to be reversed in the D2 and/or D3 dimension. If prior experience indicates that such a situation prevails for your data, you may specify that data vectors be reversed in D2 and/or D3 as a part of the ND processing. You may also delay such correction and use the Process/Reverse Matrix menu item at a later time.
Extract Half Spectrum	Extract a region of the D1 dimension for further processing. You can specify the left half of the spectrum, the right half of the spectrum, or the region to be extracted (available only for the D1 dimension).
Output Level	The amount of information written to the text window. The Quiet option prints a minimal amount. The Verbose option prints information on which vector is being processed so you can follow the progress of the calculation, and slightly lengthens processing time.

The Interactive Processing section of this control panel contains a set of controls for interactively processing the data while varying the parameters for a specific processing action. The bottom section of the control panel specifies the manner in which the processing is done. Clicking Apply performs the operations specified in the list of processing operations and redisplays the main processing panel.Selecting OK performs the indicated processing options and closes the panel.

To use this control panel, it is a good idea to set up the basic processing first. For example, try setting FT, Set Phase Correct, and Real Time to on. Then click Apply. This performs the FT operation and starts the real-time phase interface. At this point you can adjust the phase interactively. When you finish phasing (click Keep in the phase panel) you are brought back to the main 1D processing control panel, where you can select other specific processing options. Here it is often a good idea to set the Phase Correct mode to Use Current. The current phasing parameters are then used for subsequent phasing operations.

Clicking one of the Interactive Processing buttons allows you to interactively vary the parameters for the chosen processing step. The Interactive Processing options allow you to control the various processing parameters with sliders. As you adjust the sliders you can simultaneously see the effect on the FID. These interactive options also allow you to display the transformed spectrum. This way you can see the effect of varying a processing parameter on both the FID and the transformed spectrum. Clicking Keep in one of these interactive processing panels brings you back to the main control panel. To exit the main processing panel select OK or Cancel.

Process/2D Data Processing

The Process/2D Data Processing menu item (<Alt>-s2) allows you to process 2D data. Selecting this menu item opens a control panel where you can specify the Filter Type for the kind of data you want to process. In general, if this is the initial processing of raw spectrometer data set you should specify All Files (Bruker, Varian, Jeol:*) as the Filter Type. You can then navigate through the desired directories to get to the data. You must select the actual spectrometer datafile, which is usually an FID or SER file. To select the FID you can double-click the filename or click the filename and then select OK.

The next control panel presents you with the 2D header menu parameters. These parameters are taken from the header information of the spectrometer datafile. Be certain that the proper parameter values are displayed and correct them if necessary. Then set the Data Source to correspond to the type of data you have. The choices are: General, Bruker, Varian, and Jeol. The Data Source setting determines how you enter the acquisition information in relationship to how the data were collected.

General Processing

If you set the Data Source to General, you are presented with a general control panel for processing 2D data that are not related to any specific spectrometer type. This method of specifying processing parameters is similar to that used in previous versions of Felix. If you select General or Jeol processing, you are presented with the same control panel of generalized processing choices. The Data Type parameter relates to the form of the FID in the D1 (t2) dimension. The FID in the acquisition dimension is almost always Complex.

When Felix processes 2D data, the method used to collect the data in the indirectly detected dimension affects how the data are processed in the acquisition dimension. This simplifies data processing in the indirect dimension and makes it simpler for you to examine the data along this dimension. The indirect dimension is almost always actually processed as States or TPPI. Thus, for the Acquisition Mode, this parameter determines how the D1 (t2) dimension is processed, but is set based on how the indirectly detected dimension was collected. So if data were collected as a Bruker style echo/anti-echo experiment in D2 (t1), the Acquisition Mode is set to Echo/Anti-Echo and the Acquisition in D2 parameter is set to States. This is because the Echo/Anti-Echo experiment results in a complex interferogram along t1.

Bruker Processing

If you are processing Bruker data, it is generally easier to set the Data Source parameter to Bruker. This brings up a simplified control panel specifically for Bruker data. The Data Type parameter is Complex unless you have an FID where the data points are sampled sequentially. The Acquisition in D2 parameter is set based on the Bruker mode of data collection. You can generally determine this from the value of the MC2 parameter in the Bruker proc2s parameter file.

Varian Processing

If you are processing Varian data it is generally easier to set the Data Source parameter to Varian. This brings up a simplified control panel specifically for Varian data. The Data Type parameter is Complex. The Acquisition Mode parameter is set based on the type of experiment you have. Varian data is most often collected as States. If you have a sensitivity-enhanced sequence of the Lewis Kay type, then you should set Acquisition Mode to Sensitivity Enhanced. If you use the grad_sort_nd program to pre-process your data before processing on the Varian, then your data is of the Sensitivity Enhanced Type.

After you specify the acquisition parameters, select OK. The main 2D control panel then appears and allows you to specify the exact sequence of options to use during processing. See Table  20 for more information on the various processing options. When you select OK from this control panel you are asked to supply any needed additional information, and then processing is performed. When you process the D1 dimension of a 2D data set the matrix is first built and then the individual vectors from the input data set are read in, processed, and stored in the matrix.

To process the second dimension of a 2D data set, select the Process/2D Data Processing menu item again and specify a file type of Felix Matrix. In the header menu verify that the Data Source parameter is still correct. Then select the D2 dimension for processing in the main 2D control panel. When you select OK, the individual vectors from the matrix are read in one at a time, processed, and stored in the matrix.

Process/3D Data Processing

The Process/3D Data Processing menu item (<Alt>-s3) allows you to process 3D data. Selecting this menu item opens a control panel in which you can specify the Filter Type for the kind of data you want to process. In general, if this is the initial processing of raw spectrometer data, you should specify All Files (Bruker, Varian, Jeol:*) for the Filter Type. You can then navigate through the desired directories to get to the data. You must select the actual spectrometer datafile. This is usually an FID or SER file. To select the FID, double-click the filename or click the filename and then select OK.

At this point you are presented with the 3D header menu parameters. These parameters are taken from the header information of the spectrometer datafile. Verify that the proper parameter values are displayed and correct them if necessary. Set the Data Source to correspond to the type of data you have. The choices are General, Bruker, Varian, and Jeol. The setting for Data Source determines how you enter the acquisition information in relationship to how the data were collected.

General Processing

If you select General as the Data Source you are presented with a general control panel for processing 3D data that are not related to any specific spectrometer type. This method of specifying processing parameters is similar to that used in previous versions of Felix. Selecting General or Jeol processing opens the same control panel of generalized processing choices. The Data Type parameter relates to the form of the FID in the D1 (t3) dimension. The fid in the acquisition dimension is almost always Complex.

When you process 3D data in Felix, the method used to collect the data in the indirectly detected dimensions also affects how the data are processed in the acquisition dimension. This simplifies data processing in the indirect dimensions and makes it simpler for you to examine the data along these dimensions. The indirect dimensions are almost always actually processed as States or TPPI. Thus for the Acquisition Mode this parameter determines how the D1 (t3) dimension is processed but is set based on how the indirectly detected dimensions were collected. So if the data were collected as a Bruker style echo/anti-echo experiment in D3 (t1) and states-TPPI in D2(t2), then Acquisition Mode is set to Echo/Anti-Echo States-TPPI and the Acquisition in D2 and Acquisition in D3 parameters are both set to States. This is because the echo/anti-echo experiment results in a complex interferogram along t1.

The Acquisition Method determines how the t1-t2 time point values are collected. If the data are collected as a group of four complex FID's corresponding to each t1-t2 time point, then the Acquisition Method is Quartets. If the data are collected, for example, as pairs of real and imaginary t2 components for all of the real t1 values, followed by the series of pairs for the imaginary t1 values, then this is referred to as Planes. Bruker generally collects data as planes, while Varian generally collects data as quartets.

The First Incremented parameter specifies the order in which the FID's were collected. If First Incremented is set to t2, this means that the t2 parameter was incremented first and then the t1 parameter; that is, for each t1 time increment the entire set of t2 time increments is collected before proceeding to the next t1 time value. Standard Bruker sequences, which are collected as "3-1-2", should have First Incremented set to t1 because the data are collected as a series of t3-t1 planes. Bruker data collected as "3-2-1" should have First Incremented set to t2. Standard Varian data, which are normally collected as "d3, d2", should have First Incremented set to t2 because the data are collected as a series of t3-t2 planes. Varian sensitivity-enhanced sequences, which are collected as "d3, d2", should have First Incremented set to t1.

Quartet Order Parameter

The Quartet Order parameter is used when the Acquisition Type has been set to Quartets. It determines the order in which the individual elements of the complex quartet are collected and therefore which dimension in the quartet is incremented first.

If the Quartet Order parameter is set to t2, this implies that the FID's were collected in the following sequence:

FID#	t1(D3)	t2(D2)
1	real	real
2	real	imaginary
3	imaginary	real
4	imaginary	imaginary

If the Quartet Order parameter is set to t1, this implies that the FID's were collected in the following sequence:

FID#	t1(D3)	t2(D2)
1	real	real
2	imaginary	real
3	real	imaginary
4	imaginary	imaginary

Bruker Processing

If you are processing Bruker data it is generally easier to set the Data Source parameter to Bruker. This opens a control panel specifically for Bruker data. The Data Type parameter is Complex unless you have an FID in which the data points are sampled sequentially. The Acquisition in D2 and Acquisition in D3 parameters are set based on the Bruker mode of data collection. You can generally determine this from the values of the MC2 parameters in the Bruker proc2s and proc3s parameter files.

The Acquisition Order parameter is determined by which dimension (t1 or t2) is incremented first. If t1 is incremented first, then the Acquisition Order is set to "3-1-2". If t2 is incremented first then it is set to "3-2-1".

Varian Processing

If you are processing Varian data it is generally easiest to set the Data Source parameter to Varian. This opens a control panel specifically for Varian data. The Data Type parameter should be set to Complex. The Acquisition Mode parameter is set based on the type of experiment. Varian data are most often collected as States. If you have a sensitivity-enhanced sequence of the Lewis Kay type, then set the Acquisition Mode parameter to Sensitivity Enhanced. If your data were pre-processed using the grad_sort_nd program before processing on the Varian, then your data are of the Sensitivity Enhanced Type.

The First Incremented parameter is set based on which dimension is incremented first. Varian data are most often collected as "d3, d2". The Quartet Order parameter is set based on the order in which the elements of the complex quartet of FID's were collected. This parameter is set based on the array parameter in the Varian procpar file. Set this parameter to phase, phase2, or phase2, phase, depending on the value of the array parameter.

After you specify the acquisition parameters, select OK. You should now see the main 3D processing control panel, which allows you to specify the exact sequence of options that will be used during processing. See Table  20 for more information on the various processing options. When you select OK in this control panel, you are asked to supply any needed additional information and then processing is performed. When you process the D1 dimension of a 3D data, set the matrix is first built and then the individual vectors from the input data set are read in, processed, and stored in the matrix.

To process the second dimension of a 3D data set, select the Process/3D Data Processing menu item again. Now specify a file type of Felix Matrix. In the header menu, be sure that the Data Source parameter is still correct. Then in the main 3D processing control panel, select the D2 dimension for processing. When you select OK, the individual vectors from the matrix are read in one at a time, processed, and stored in the matrix.

Process/3D Plane Processing

The Process/3D Plane Processing menu item (<Alt>-sl) allows you to process a 2D plane from a 3D data set. This menu item is similar to the Process/3D Data Processing menu item described above. The difference is that plane processing creates a 2D matrix instead of a 3D matrix. This function builds the 2D matrix, reads in each vector from the input data set, processes them, and stores them in the 2D matrix. After performing this step, you will have a 2D matrix that is processed in the D1 dimension only. You must then use the Process/2D Data Processing menu item to process the D2 dimension of this new matrix.

In the control panel for processing parameters you can specify a D1-D2 (t3-t2) plane or a D1-D3 (t3-t1) plane for processing. The other options in the various panels are the same as for 3D processing, described above.

Process/4D Data Processing

The Process/4D Data Processing menu item (<Alt>-s4) allows you to process 4D data. Selecting the menu item opens a control panel in which you can specify the Filter Type for the kind of data you want to process. In general if this is the initial processing of a raw spectrometer data set, you should specify All Files (Bruker, Varian, Jeol:*) as the Filter Type. You can then navigate through the desired directories to get to the data. You want to select the actual spectrometer datafile. This is usually an FID or SER file. To select the FID you can double-click the filename or click the filename and then select OK.

Next you are presented with the 4D header menu parameters. These parameters are taken from the header information in the spectrometer datafile. Verify that the proper parameter values are displayed and correct them if required. Set the Data Source parameter to correspond to the type of data you have. The choices are Unknown, Bruker, Varian, and Jeol. 4D data are handled with a general processing scheme that is not specific to any spectrometer type. The acquisition parameters to be entered for 4D data are analogous to those for 3D data with an additional dimension. For more information on entering the acquisition information see General Processing. The processing options are the same as those for 2D and 3D data.

Process/Phase Correct Matrix

Processed data occasionally require re-phasing in one or more dimensions. The Process/Phase Correct Matrix (<Alt>-ph) menu item allows you to re-phase a previously processed ND dataset that is currently open in the frame. The control panel allows you to specify the dimension to rephase and the method. In addition to automatic phasing, you can give explicit values for the phasing parameters Phase0 and Phase1 or adjust the phase parameters interactively. Selecting OK re-phases all vectors in the matrix, using these phase parameters (the matrix must be write-enabled).

If you select the automatic phasing function, you can select PAMPAS or APSL as the phase-detection algorithm and can define some excluded areas, to exclude noise while searching the sample peaks for calculation of phase parameters. You can also specify a Filter Width, which is the minimum peak width (at the peak bottom) required for sample peaks. Felix automatically suggests a value for Filter Width based on the matrix data, but you can change the value.

Process/Baseline Correct Matrix

Felix provides a host of baseline-correction options, and three of the more popular methods, FLT, convolution, and FaceLift, have been used as the basis for a set of post-transform baseline-correction menu items. All the available baseline-correction methods are discussed in detail under Process/Baseline correction. In the Process/Baseline Correct Matrix menu item (<Alt>-pb), you need to specify the method and which dimension the baseline correction should proceed along.

FLATT method

The FLATT menu item conducts baseline correction on the selected dimension of the transformed ND data using the algorithm of Guntert and Wuthrich (1992). The FLATT algorithm automatically discriminates baseline segments and uses a linear least-squares solution to fit a trigonometric series to the baseline points. The Baseline width determines a minimum chi-square value. A value that represents the half-width (in points) of the broadest resonance in the spectrum generally yields satisfactory results. Points to Correct specifies the number of trigonometric terms to use in the baseline fit, and Tau specifies the factor by which a segment may exceed the minimum chi-square value and still be considered a baseline segment. You may directly specify the source of the chi-square value or allow the utility to derive it (the value then represents the average over all vectors in the matrix). Armed with these values, the function loads and baseline-corrects every vector in the matrix (the matrix must be write-enabled).

Convolution method

The Convolution menu item conducts baseline correction on the selected dimension of transformed ND data using the algorithm of Dietrich et al. (1991) as developed by W. Massefski. This ABL-based utility automatically discriminates baseline segments and conducts a running-average convolution of the baseline points, while a simple linear correction is applied to the intervening spectrum regions. The Noise size specifies the convolution width (in points) for the baseline regions, and the Peak size represents the half-width (in points) of the broadest resonance in the spectrum. Using these values, the function loads and baseline-corrects every vector in the matrix (the matrix must be write-enabled).

FaceLift method

The FaceLift function conducts baseline correction on the selected dimension of transformed ND data using the algorithm of Chylla and Markley (1993). This signal recognition-based utility automatically identifies baseline points and filters the high-frequency noise along other dimensions before the baseline matrix is subtracted from the original matrix. The Filter Width is the half-width of the smoothing data window over which datapoints are sampled. The recommended range is 32-64 data points (powers of 2 are not necessary). The Number of Standard Deviations is used to determine a threshold standard deviation, above which any point is considered to be a signal point. The recommended range is 2.5-3.0. The D1 (D2, D3, or D4) Points to Smooth is the half-width of the smoothing data window that is used to smooth the base-point correction matrix along D1 (D2, D3, or D4). If it is the same dimension as that being baseline corrected, you should use the same value as for the Filter Width. Otherwise, a value of 2-4 is recommended (the matrix must be write-enabled).

Process/Reverse Matrix

Certain hypercomplex phase-cycling protocols effectively render the complex-conjugate of what is normally expected by the ND States processing utility. Such data appear to be reversed in the D2 and/or D3 and/or D4 dimension. If prior experience indicates that such a situation prevails for your data, you may specify that data vectors be reversed in D2 and/or D3 and/or D4 as a part of the ND processing. The Process/Reverse Matrix (<Alt>-pr) menu item allows you to specify the dimension to reverse in the currently open matrix. Every vector along the specified dimension is reversed (the matrix must be write-enabled).

Process/Utilities

Process/Utilities/Squeeze Matrix

The Process/Utilities/Squeeze Matrix (<Alt>-pss) menu item allows you to squeeze the current matrix (that is, to discard all the points below the threshold you define). This is useful for retaining only those portions of the matrix where real peaks can be found. Depending on the threshold, you can compress the file quite a bit, which can speed up the redraw and shorten the access time for originally large 3D and 4D spectra.

Caution

This procedure is irreversible, and some actions (for example, volume measurement and peak optimization) may not work well on such a matrix.

Process/Utilities/Unsqueeze Matrix

The Process/Utilities/Unsqueeze Matrix (<Alt>-psu) menu item creates an unsqueezed matrix from a previously squeezed one. This menu item is not the reverse of Process/Utilities/Squeeze Matrix, since it merely inserts zeros in those places in the matrix that were previously (during a squeeze) discarded.

Process/Utilities/Transpose Matrix

The Process/Utilities/Transpose Matrix (<Alt>-pst) menu item allows you to swap two dimensions of a processed matrix.

Process/Utilities/Projection

The Process/Utilities/Projection (<Alt>-psp) menu item allows you to create a 2D projection of the current 3D or 4D matrix.

Process/Utilities/Diagonal Plane

The Process/Utilities/Diagonal Plane (<Alt>-psd) menu item allows you to extract a 2D diagonal plane from a typically homonuclear 3D matrix.