Insight II



9       Analysis Module


Implementation

The Analysis module is used to display data, especially from molecular conformation analyses, in graphs or molecular animation. Typically, the input data of molecular conformations comes from the Discover program (separately licensed from Biosym), but this data can come from any source, provided it is formatted in a way that is recognizable to the Insight II program.

Three types of graphs can be created with commands in the Analysis module. The first type of graph, usually referred to as a simple trajectory graph, uses atomic trajectories as its most basic data source. Properties of the trajectories, such as atomic distances or angles, are displayed as curves. The second type of graph, called a cluster graph, also uses atomic trajectories. Its purpose is to display the relationship between each conformer and every other conformer. This relationship is expressed as an RMS value, and the graph is a cluster of unconnected points that are color coded to the RMS level, thereby indicating families of related conformations. The third type of graph, sometimes called an inanimate graph, does not use trajectory data. Instead, it uses data from simple ASCII files that you may create on your own or by using the Put Graph command. In any event, these ASCII files used to create inanimate graphs contain columns of numbers which can be plotted against one another to form curves similar to those displayed in simple trajectory graphs.

Trajectory data from the Discover program's history and archive files (refer to the Discover documentation appendices for a complete description of an archive file) can be analyzed quantitatively by defining properties of interest, identifying the atoms that uniquely define the property, and graphing those properties against each other. The data can be qualitatively analyzed by simply watching an animation of the molecule or molecular system as it moves through its conformations in real time. When a molecule is animated, each atom follows the path prescribed by the trajectory data. For both qualitative and quantitative analyses, the trajectory data can be filtered to remove unwanted vibrational frequencies.

Simple trajectory graphs have a cursor that moves along the plot in exact synchronization with the animation and dynamic data labels that give the coordinates of the cursor. In this way, the cursor highlights the data point on the curve that corresponds to the current conformation being shown in the animation. Animations can be temporarily frozen, and specific conformations corresponding to specific plot points can be viewed. (Note that although inanimate graphs may have curves which look like those in simple trajectory graphs, they are not necessarily derived from trajectories and are, therefore, unchanged by animations.)

Archive files containing conformers derived from NMR or SCS (Systematic Conformational Search) can also be used in the same way that Discover program data is, to create graphs, etc.

Six types of properties are available for analysis when creating simple trajectory graphs: time, frame number, temperature, energy, distance (including point-plane), and periodic (angles, dihedral angles, and plane-plane angles). Multiple properties may be plotted on the same graph, as long as the properties have the same units of measurement. Properties can be defined in any order. There is no software limit on the number of properties that can be defined, or the number of graphs that may be displayed. However, more than six graphs or six plots on a single graph generally results in a cluttered display.

If the trajectory data is read from a history file, the component energies and the time step between frames are read from the history file; all other properties are calculated by the Insight II program. If the data is read from an archive file, the total energy is read from the file, and the time step is taken to be 1 fs (femtosecond) per frame within the data file. All other properties are calculated by the Insight II program.

For those commands that use the loaded trajectory information (such as Animate, and cluster and trajectory graph construction commands), frame one (1) is the first frame loaded. This may not be the first frame contained within the history or archive data file. Additionally, the loaded trajectory data frame numbers run consecutively, starting from one (1) and ending with the maximum frame number without skipping numbers. These frame numbers may not exactly match the frames that were loaded from the data file, but there is a one-to-one correspondence between them. For example, the first frame loaded from the data file may be frame five (5), but it is referred to as frame one (1) once it has been loaded. The second frame loaded could be frame ten (10), but it is referred to as frame two (2) after it is loaded.

The calculation of distances, angles, and dihedral angles is straightforward, and is not presented here. However, the calculation of a plane that best fits more than three points is more complex and is presented in the next section.

Cluster graphs have no cursors or dynamic data labels. Instead, they have legends that describe the color coding used when displaying clusters of points representing related families of conformers.

Inanimate graphs (i.e., those created from simple columns of numbers) are not linked to any molecule and do not have cursors, dynamic data labels, or legends. However, like all three kinds of graphs, they do have axis labels to describe what is being plotted, and axis min/max labels showing the numeric bounds of the data being displayed.

When new trajectory data is loaded, either from a different trajectory file or by using a combination of frames differing from the presently loaded trajectory data, any non-animate graph is deleted. In other words, any graph that is linked to the molecule(s) associated with the present trajectory data disappears. This is necessitated by the correspondence between the trajectory data and each graph data point.

Once a graph is defined it can be manipulated in much the same way as any other object within the Insight II program. Graphs may be scaled, moved, rotated, blanked, or unblanked.

Average Plane Evaluation

A plane P is described in general with the following equation:

Eq. 9¯1             P(X) : ax + by + cz + d = 0 with x = (x, y, z)

However, this description has an infinite set of a, b, c, and d, so that when they are scaled, the same plane remains. For a unique representation of this plane. From this equation, three unique equations can be derived:

Eq. 9¯2             Qz : z = f(x, y) or z = 13x = m1y + n1 or

Eq. 9¯3             Qx : x = g(y, z) or x = 13y = m2z + n2 or

Eq. 9¯4             Qy : y = g(x, z) or x = 13x = m3z + n3

Obviously, Eq. 9-2 is not appropriate to describe the plane x = 0. In this case, Eq. 9-3 is the best representation.

To derive the average plane from a set of N points, it is important to get the best plane given a merit function which usually is:

minimize (a, b, c, d) by using:

This is the familiar least square fitting function. This same function is used in chi-square statistics, the so-called maximum likelihood estimator (Pattabiraman et al. 1985).

Given a set of (a, b, c, d), certain questions arise, namely, "How good is the fit? Is there a better fit?" When the average plane is evaluated, the desired information includes:

1. The parameters (a, b, c, d).

2. An error estimation for these parameters.

3. A statistical measure of the fit.

The next lines describe how the most likely plane is obtained. To derive the formula, take the plane equation:

Eq. 9¯5            

The chi-square value estimates how well the model agrees with the data (our points). Take the first derivative for this merit function for the l, m, and n unknowns. Thus, you get:

with S = N points, Su = and Suv = with u and v equal to x, y, and z.

By solving this three-dimensional system you come up with:

with duv = S suv - Su Sv and d = dxx - dyy - dxy dxy and with S, Sx, Sy, Sz defined as before.

From these l, m, and n values we obtain the a, b, c, d coefficient of Eq. 9-1 (Press et al. 1986) with:

Frequency Filtering

The Filter Trajectory command extracts motions that correspond to interesting vibrational frequencies from a set of atomic trajectories.

The Discover program can be used to simulate a set of atomic trajectories. The molecule(s) can then be displayed and animated using the Insight II program so that it moves through the path of the simulated trajectory.

Such simulations may cover a wide variety of molecular vibrations. So, a molecule can display the relatively quick movements of bond stretches, scissors, wags, etc. It can also show the slower movements such as linear chains rotating, bending, and twisting.

Viewing all of these movements at once may make it difficult to focus on one or more particular kinds of vibrations.

A Fast Fourier Transform (FFT) can be used to convert the trajectory data from the time domain to the frequency domain. Those frequencies not corresponding to interesting vibrations can be filtered out. This filtered frequency data can then be transformed back into the time domain. When animated using this filtered trajectory data, the molecule displays only those vibrations that are of interest.

Prior to using the Filter Trajectory command, a trajectory data set must be loaded (using the Get Trajectory command). After loading this data, the Insight II program reports the overall number of frames that exist. This number can be important when you decide what values to give to the Filter Trajectory command, so it should be noted.

The trajectory data set is modified by the Filter Trajectory command. To get back to the original trajectory data, re-execute the Get Trajectory command.

As described below, the Filter Trajectory command is limited to modifying a number of frames that is an exact power of 2. For example, 1024 or 2048 or 4096 frames can be modified, but some arbitrary number, such as 1000 or 2000 frames, cannot. The number of frames that is modified is calculated by subtracting the starting frame number (which is an input to the Filter Trajectory command) from the total number of available frames. If this is an even power of 2, then this is how many frames are modified. Otherwise, the number of frames modified is less than this total number of frames available and is the nearest power of 2.

This means that the Filter Trajectory command might leave some of the frames in the trajectory set untouched. You must be aware of this and use that knowledge when you specify start/end/step values for animation or when you create graphs.

The information about what frames have been touched by the Filter Trajectory command versus the frames that are left intact is reported to the textport during the execution of the command. These numbers should be noted for future reference.

Note that the limitation of modifying only a number of frames that equals a power of 2 is not arbitrary. With very few exceptions, Fourier transforms and their inverses can only be performed on computers using some variation of recursion with a binary tree. In other words, they work with arrays of data that have a number of elements that equals a power of two. Arrays of arbitrary length can be handled in one of two ways. The array can be clipped down to a size that is a power of two. This is the technique that is used with the Filter Trajectory command. Or, the array can be extended to a size that is a power of two. When this is done, each newly added element of the array must be given some value. There is no good rule on how this padding should be done, although there are references in various published articles that suggest that padding with zeros or symmetric reflections of the original data have been attempted. Without exception, such padding introduces fictitious data that results in frequencies that are not truly present in the original data. Bracewell (1990) recently stated, "When sampled input is not naturally zero outside the sampled range, packing with zeros introduces unwarranted discontinuities whose effects on the transform, such as overshoot and negative-going oscillation, may be undesirable. Packing with plausible but unobserved data can eliminate the undesired artifacts and is probably practiced in more cases than are admitted to. Investigators often mitigate the effects of discontinuities in the data by multiplying by a tapering factor; they should then explain that they value freedom from negatives more than accuracy of amplitude values of spectral peaks or than resolution of adjacent peaks."

One of the parameters in the Filter Trajectory command indicates which frame to begin processing with. It specifies where, in the total set of trajectory data, the FFT filtering is to begin. In most cases, the starting frame number should be one (1), meaning that the FFT filtering should begin with the first frame in the trajectory set.

It is possible to use a value other than one in order to see the difference between the original molecule motions and those that have been filtered. Also, you might want a to use a number greater than one if you want to skip frames in the beginning of the trajectory set that correspond to the molecule reaching temperature equilibrium.

There is another, more subtle reason why you might want the starting frame number to be a value other than one. As described previously, the FFT algorithm can only process a number of frames that is equal to an exact power of 2. Even if you request that the processing begin with the first frame, the Filter Trajectory command may not process every frame. Processing would start with the first frame, but may end before the end of the trajectory data set if the total number of frames in the set is not an exact power of 2.

Consider the case where the trajectory data set has 1500 frames. If the starting frame number is one, then the first 1024 frames are processed and the remaining 476 frames are left untouched.

By setting the starting frame number to 476, the untouched portion of the trajectory set is, in essence, moved from the end of the set to the beginning of the set.

Regardless of the value that you set for the starting frame number, the Filter Trajectory command might leave portions of the trajectory set untouched because of the requirement of processing only a number of frames that is an exact power of two.

Trajectory data sets that are only partially filtered can lead to discontinuities when you cross the boundary between filtered and unfiltered frames when using the Animate Trajectory or Construct_Graph Trajectory commands. The effects are not wrong, but they may be startling because unfiltered and filtered trajectories can have very different characteristics.

Another parameter in the Filter Trajectory command specifies the time step between frames (in femtoseconds). This value can only be determined if you know how the trajectory data was calculated. By default, dynamics trajectories calculated using the Discover program (through the Insight II program) have a time step of 10 fs. This is because, while new conformers are, by default, calculated every femtosecond, only every tenth conformer is put into the history file. Thus, the time step between frames is, by default, 10.

Both the time between conformers and the number of conformers skipped between output to the history file can be adjusted before the simulation is run.

The time between frames in a history file may be calculated by multiplying the time step by the frequency of writing to the history file. In terms of the input file (.inp) and output file (.out), these parameters are controlled with the "steps" and "write history" qualifiers to the initialize dynamics command.

A basic rule of the sampling theory that governs discrete Fourier transforms states that the input data must be sampled at least twice as fast as the fastest frequency in the system. Two times the maximum frequency corresponds to what is called the Nyquist sampling rate. In theory, data collected with a sampling rate that is too slow introduces some errors in the Fourier transform. It is often difficult or impossible to predict exactly how severe these errors will be.

In terms of using the Filter Trajectory command, the Nyquist limitation means that it is best to run simulations of hydrocarbons such that the time between frames that are captured in the history (or archive) file is not more than 5 fs. If the time between frames is greater, you may see some erroneous movements in the molecule. (However, practical experience has shown that these errors are usually negligible as far as the display is concerned and the resulting animations look fine.)

The recommendation to capture frames every 5 fs is based on the idea that the C-H bond stretch is one of the fastest (if not the fastest) vibration in a hydrocarbon simulation. Its period is about 11 fs. Half of this is about 5 fs.

Note, however, that the Filter Trajectory command allows any positive value to be used when specifying the time between frames. It is up to you to make sure that it is correct and that it is not so large as to cause problems.

The final piece of information needed for the Filter Trajectory command is the range of frequencies to retain after the filtering. The units of frequency used by the Insight II program are reciprocal centimeters. In other words, frequencies are referred to by wave numbers.

The function of the Filter Trajectory command is to remove from a set of trajectory data all but the interesting motions. Such motions are identified by their vibrational frequency (range).

The zero frequency corresponds to the motionless molecule in its average conformation. This is a simple positional average over a set of positions. It is important to note that this kind of averaging can result in atomic positions that do not fall on the trajectory path. For example, atoms that sweep circles in space have an average position that is in the center of the circle, not somewhere on the circle itself. For this reason, some atom groups might look distorted. Methyl groups, for example, might show hydrogens that are collapsed toward that central point (over the carbon). Nonetheless, the overall topography of the molecule is generally maintained.

Frequencies of only a few hundred reciprocal centimeters generally correspond to global motions in the molecule, such as the rotation or bending of linear chains. Higher frequencies generally relate to bond stretches, or wags or scissor movements, and so on. These are, of course, generalizations. There are many notable exceptions. For example, the bond stretches of iodine correspond to frequencies of only a couple hundred reciprocal centimeters.

The frequency range from zero to zero is always included and does not need to be explicitly given.

Beyond exploring the more subtle aspects of using the Filter Trajectory command, it is useful to provide some detail of the mathematics that allow this filtering to take place. As mentioned above, the filtering algorithm is straightforward:

1.   The position of an atom in a simulated trajectory can be regarded as a function of time (p = f(t)).

2.   A discrete Fourier transform brings data from the time domain to the frequency domain by

where H(v) is the frequency function corresponding to f(t), the position function, i is the imaginary unit, v is the frequency, t is time, N is the number of discrete data points, and is pi.

3.   Unwanted frequencies are zeroed out in the resulting array of calculated frequencies.

4.   The filtered frequency array is brought back into the time domain by

where f(t) is the position function, H(v) is the frequency function corresponding to f(t), i is the imaginary unit, v is the frequency, t is time, N is the number of discrete data points, and is pi.

Additional Reading on Frequency Filtering

It is beyond the scope of this document to provide a complete bibliography of books and periodicals that describe the theory and application of Fourier transforms and filtering. For a general treatment of numerical transforms with some discussion of caveats of applications, see Bracewell (1990); for a cookbook approach to model computational techniques for Fourier transforms, refer to Press (1986); and for an application of Fourier transformations to filter atomic trajectories, see Sessions (1989).

Clustering of Conformations into Families

The atomic trajectory data contained in a history or archive file created by the
Discover program can be thought of as a set of conformers for a molecule or system of molecules. It is often important to examine whether families of conformers with closely related topography exist. For example, each such family may correspond to conformers surrounding a particular energy well in the simulation.

The analysis of sets of conformers to reveal such families is performed by creating a cluster graph. Such a graph looks like a box filled with unconnected points. Each point represents how a particular conformer compares with another conformer. Each point is color coded to indicate how well it compares. (The z axis of the graph shows this comparison value numerically as well.)

Cluster graphs are created by setting the Cluster_Graph parameter on in the Construct_Graph Trajectory command. The exact steps that you should follow to create a cluster graph are more fully described later in this chapter. An example of a cluster graph creation is included in the tutorial for this chapter.

The topographical comparison that is done between conformers to classify families is based on determining a minimum RMS value from superimposing selected atoms. Essentially, this procedure is the same as the superimposition (with RMS calculation) done in the Superimpose command in the Transform pulldown. Here, too, the basic equation used to calculate the RMS is given as:

where the superimposition is aimed at aligning N atoms in two conformers of the same molecule and xi, yi, zi represent the spatial coordinates of atom i in one conformer while xi¢, yi¢, zi¢ represent the spatial coordinates of the same atom in the other conformer.

Mean Square Displacement Precision

For n discrete frames in a trajectory, the MSD, <[ri(t) - ri(0)]2> of frame from frame 0 is actually calculated by:

Where max = (n - ).

The short-time MSD, with a small , may be determined with slightly greater statistical precision because the number of terms in the average max may be greater.

This calculation is workable for both the minimum image and the explicit image options for periodic boundary conditions in the Discover program. A correction is made for (translational) switching between images, if the explicit image model in Discover is used in the simulation.

Radial Distribution Functions

A radial (or pair) distribution function (RDF), , describes the probability of finding an atom of type b at a distance between r and from atoms of type a as a function of the a - b separation r. The radial distribution function calculations were initially applied to investigate the modification of the solvent structure induced by the presence of the solute. Recently, the RDF has been used to study the dynamical structural modification of dense systems during a molecular dynamics simulation (Avbelj et al., to be submitted). In addition, the RDF can serve broader purposes, with weight factors for each atom at distance r. One such example is the hydrophobicity contrast function which has been used to search metal binding sites in proteins (Yamashita et al. 1990).

Furthermore, the ensemble average of any pair function, such as the nonbonded energy or the pressure, can be related to (Allen and Tildesley 1989).

Eq. 9¯6            

where is the average number of type b atoms at the distance between r and from a type a atom, is the bulk density of type b atoms, and is the average local density of type b atoms in the shell volume, , between r and from a type a atom.

Note that the type a atoms refer to the atoms specified by Molecule Spec, and the type b atoms by Molecule 2 Spec.

The divisor, therefore, normalizes the distribution so that when is the same as the bulk density. Thus, the RDF represents the factor of local density/bulk density for type b atoms around type a atoms. Note that the bulk density can either be calculated as for a system with a cell, (the Bulk_Density parameter is on) or as a unit for a molecular system with no cell (the Bulk_Density parameter is off).

The coordination number, , represents the number of type b atoms around a type a atom at a distance R. The average coordination number is expressed as:

Eq. 9¯7             .

In this calculation, the is also partitioned into intermolecular and intramolecular contributions. To explore the structure of a system during a dynamics simulation, options are provided for calculating either for an individual configuration, a set of configurations (trajectory frames), or the average over a set of frames. When calculated individually for each of the configurations, they can be plotted as a 3D graph, .


Command Summary

In addition to the core pulldowns in the top menu bar, the Analysis module adds five pulldowns, Graph, Trajectory, Pseudo_Atom, Spectra, and Background_Job to the lower menu bar.

Graph Pulldown

The Graph pulldown allows you to create and manipulate graphs. These commands are described in the following chapter, Graph Pulldown.

Trajectory Pulldown

The Trajectory pulldown contains commands for working with trajectory data. These commands use information from previously run molecular mechanics simulations.

Some of the Trajectory pulldown commands are colored gray (e.g., the Cluster command, and the commands ending with _Def). These are not selectable from the pulldown because they are activated automatically, in a specific sequence, by other non-grayed commands in this pulldown. However, all the commands shown in the pulldown are typable.

Get Trajectory

The Get Trajectory command is used to load into the Insight II program atomic trajectory information contained in history or archive files that have been generated in previous runs of the Discover program. The command sets up information needed by other commands such as Construct_Graph or Animate.

Put Trajectory

The Put Trajectory pulldown allows you to create an archive file from a loaded trajectory. It may contain all or just some of the frames of trajectory data that have been loaded. These frames may be reordered as they are written out.

The original trajectory data may come from a variety of sources: an archive file, a history file, or a torsion file.

Filter Trajectory

The Filter Trajectory command allows you to filter trajectory sets so that only motions with interesting vibration frequencies remain. For details on the method used see the previous section on Frequency Filtering.

Animate Trajectory

The Animate Trajectory command is used to make molecules move through the conformations calculated during the simulation, as read from the contents of a Discover history or archive file (trajectory file).

Unanimate Trajectory

The Unanimate Trajectory command is used to permanently stop an existing animation. The display of the motion of the molecule or molecular system stops and related printouts are cleared from the screen. This is different from temporarily stopping an animation with function key <F8>.

Conformation Trajectory

The Conformation Trajectory command allows you to examine conformations that come directly, or are derived from trajectory sets previously computed by, the Discover program. The molecule or assembly affected by this command is always whatever molecule or assembly was specified when the Get command was last executed.

Family Trajectory

The Family Trajectory command is used to create, or modify, a family of conformers based on the values determined from a cluster graph computation. Once a family has been defined it is identical to an assembly, where each member corresponds to some frame within the trajectory data. All members of a given family must be derived from cluster graphs that were defined using the same set of atoms for comparison.

Repartition_Cluster Trajectory

The Repartition_Cluster Trajectory command allows you to change the number of conformer families. You can repartition cluster graph levels, or update the existing display of a cluster graph, without having to recalculate the RMS values.

Construct_Graph Trajectory

The Construct_Graph Trajectory command is used to create graphs that plot common properties, such as energy, time, distances, and periodic properties from previous simulations. You can also add new plots to existing graphs. The relationship between any two or three properties can be plotted in any order.

Axis_Function Trajectory

The Axis_Function Trajectory command is used to associate functions, such as time and energy, with a graph axis when plots are created. This command is automatically invoked by the Construct_Graph command and is not usually entered separately.

Cluster Trajectory

The Cluster Trajectory command allows you to create a cluster graph that displays the results of a structural comparison throughout the defined trajectory data. Every frame within the trajectory specified by the Start, End, and Step parameters is compared to every other frame and the RMS resulting from the superimposition of the frame pairs is displayed within the cluster graph. This command is automatically invoked by the Construct_Graph command and is not usually entered separately.

Distance_Def Trajectory

The Distance_Def Trajectory command is used to assign a distance between two unique atoms to an axis on a graph. For example, to see how a distance between two atoms varies over time, you can assign a time function to the x axis using the Axis_Function command and a distance value to the y axis with the Distance_Def. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

COM_Distance_Def Trajectory

The COM_Distance_Def Trajectory command allows you to define a distance between two centers of mass. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

Angle_Def Trajectory

The Angle_Def Trajectory command is used to assign an angle between three unique atoms to an axis on a graph. For example, to see how an angle between three atoms varies over time, you can assign a time function to the x axis using the Axis_Function command and an angle function to the y axis with the Angle_Def command. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

Dihedral_Def Trajectory

The Dihedral_Def Trajectory command is used to assign a dihedral angle defined by four unique atoms to an axis on a graph. For example, to see how a particular dihedral angle varies over time, you can assign a time function to the x axis using the Axis_Function command, and dihedral to the y axis with the Dihedral_Def command. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

PP_Ang_3_Def Trajectory

The PP_Ang_3_Def Trajectory command is used to assign an angle between two planes to an axis on a graph, using only three points to calculate the plane. For example, if you want to see how a particular angle between two planes varies over time, you can assign time to the x axis using the Axis_Function command, and PP_Ang_3_Def to the y axis using the PP_Ang_3_Def command. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

PP_Ang_n_Def Trajectory

The PP_Ang_n_Def Trajectory command is used to assign an angle between two planes to an axis on a graph, using three or more unique points to calculate the plane. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

pP_Dist_3_Def Trajectory

The pP_Dist_3_Def Trajectory command is used to assign a distance between a point and a plane, using only three points to calculate the plane. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

pP_Dist_n_Def Trajectory

The pP_Dist_n_Def Trajectory command is used to assign a distance between a point and a plane to an axis on a graph, using three or more points to calculate the plane. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

COM_P_Dist_n_Def Trajectory

The COM_P_Dist_n_Def Trajectory command allows you to define a distance between a center of mass and a plane. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

Energy_Def Trajectory

The Energy_Def Trajectory command allows you to use the energy components found within a Discover program history file to define an axis energy function. If the trajectory information is from an archive file, then only the total energy may be defined. This command is automatically invoked by the Axis_Function command and is not usually entered separately.

Tabulate Trajectory

The Tabulate Trajectory command is used to create a table that displays the selected trajectory properties. Each property is represented by a column of information, with the name of the property above each. You can select which frames are output as well as which atoms.

Pseudo_Atom Pulldown

The Pseudo_Atom pulldown contains commands to create and modify pseudoatoms. The Define Pseudo_Atom command creates the pseudoatoms. The Rename Pseudo_Atom and Delete Pseudo_Atom commands modify the already defined pseudoatoms. The List Pseudo_Atom command lists information about the pseudoatoms. For more extensive descriptions of these commands, refer to their descriptions in the Builder Module chapter of this guide.

Background_Job Pulldown

The Background_Job pulldown allows you to set up background jobs to run concurrently or interactively with the Insight II program. You are given the choice of whether to send background jobs to a local or remote host. This pulldown is generic and is found in many modules that run background jobs. The Background_Job pulldown contains the following commands: Setup_Bkgd_Job, Control_Bkgd_Job, Completion_Status, and Kill_Bkgd_Job. Refer to the Background Job Pulldown chapter for more information.


Methodology

Quantitative analysis of trajectory data usually results in the creation of a simple trajectory graph or a cluster graph. The steps usually followed to create a simple trajectory graph are:

1.   The Get Trajectory command is used to read in trajectory data (a .his file) and associate the data with the molecule or molecular system being studied (a .car or .arc file). The associated system must already be loaded when you issue the Get Trajectory command.

2.   The Filter Trajectory may (optionally) be used to remove unwanted vibrational frequencies from the trajectory data.

3.   The Construct_Graph Trajectory command is used to begin the sequence of steps followed to create a simple trajectory graph. In this case, the Cluster_Graph parameter should be set to off. If the New_Graph parameter is set off, you may use the Graph_Name parameter to add plots to an existing graph. When New_Graph is on, a new graph is created. If the Map parameter is on, you can indicate which contour map to overlay in the graph through the Map_Name parameter.

4.   When the Construct_Graph Trajectory command is executed (from the menu), the Axis_Function Trajectory command is automatically invoked. The type of property (time, energy, temperature, or geometry) to be plotted on each axis is selected and the command is executed.

5.   For geometric or energy properties, when the Axis_Function Trajectory command is executed another command (such as Distance_Def, Angle_Def, Dihedral_Def, PP_Angle_3_Def, PP_Angle_N_Def, pP_Dist_3_Def, pP_Dist_ N_Def, Com_P_Dist_N_Def, or Energy_Def from the Trajectory pulldown) is automatically invoked. In these commands you give the appropriate details (such as atom selections or energy components) to fully define the property you have already chosen.

6.   Finally, you set the Function_Mode parameter in the Axis_Function Trajectory command to End_Graph and execute the command. This creates the graph curve.

The steps usually followed to create a cluster graph are:

1.   The Get Trajectory command is used to read in trajectory data and associate the data with the molecule or molecular system being studied.

2.   The Filter Trajectory may (optionally) be used to remove unwanted vibrational frequencies from the trajectory data.

3.   The Construct_Graph Trajectory command is used to begin the sequence of steps followed to create a simple trajectory graph. In this case, the Cluster_Graph parameter should be set to on.

4.   When the Construct_Graph Trajectory command is executed (from the menu), the Cluster command is automatically invoked. Leaving the End_Definition parameter off, a list of atoms to be used in the RMS comparisons is built by one or more executions of the Cluster Trajectory command. Then the End_Definition parameter is turned on, and the maximum RMS cutoff value and the number of families to be grouped is defined. When the Cluster Trajectory command is executed, the cluster graph is built.

The Get Graph, Equation Graph, Correlate Graph, and Boolean Graph commands can be used to create more simple trajectory graphs as well as inanimate graphs. These commands are discussed in detail in Chapter 8 of this manual.

Qualitative analysis of trajectory data may be achieved with the Animate Trajectory command to view molecular motions in realtime.


Memory Limitations

It is important to note that the Get Trajectory command causes large amounts of memory to be used for the storage of the data. If you are only interested in animating the molecule or molecular system without creating any graphs, the Get Trajectory command may be omitted. If the Get Trajectory command is not used to read in the data, the file containing the data is specified as a parameter to the Animate command. It should further be noted that the amount of memory used in the Animate command is directly proportional to the complexity of the display. If you are animating large molecules, limit the number of atoms displayed and the number of colors used in order to conserve memory. If you are only interested in creating graphs, the molecule(s) or molecular system should not be animated.


Analysis Tutorial

As of this release, most tutorials are now available online for use with the Pilot interface. To access the online tutorials for Analysis, click the mortarboard icon in the Insight II interface.

Then, from the Open Tutorial window, select Insight II tutorials, then the Analysis module and choose from the list of available lessons:

Lesson 1 Using Trajectory Commands

Lesson 2 Using Pseudoatoms

Lesson 3 Getting the Data of a Box of Water

Lesson 4 Compute the Pair Distribution Function for a Box of Water

Lesson 5 Compute the Mean Square Displacement for a Box of Water

You can access the Open Tutorial window at any time by clicking the Open File button in the lower left corner of the Pilot window.

For a more complete description of Pilot and its use, click the on-screen help button in the Pilot interface or refer to the Insight II User Guide.




Last updated December 17, 1998 at 04:27PM PST.
Copyright © 1998, Molecular Simulations Inc. All rights reserved.