| Search_Compare |
An additive amount and/or a scale factor can be applied to each van der Waals radius before the molecular volume is created. (These are applied only during volume creation and do not change the van der Waals radii referred to by other functions in Insight II.)
Eq. 1 is minimized using the Levenburg-Marquardt method (Marquardt, 1963), which is very efficient for nonlinear least-squares problems. Minimization by this method varies smoothly between the extremes of the steepest-descent and the Newton methods. The former is used far from the minimum, with the algorithm switching continuously to the latter as the minimum is approached. The Levenburg-Marquardt method requires the first derivatives of the minimizing function to be known. Here, the first derivatives with respect to all variables are calculated using the analytical solutions derived from Eq. 1.
Overlapping Molecules
You cannot superimpose pseudoatoms, since this is not supported by the current version of the Overlap functionality.
In the electrostatic calculation, Pr in Eq. 6 is replaced by a Gaussian function approximation of two terms, as in Eq. 14. The integrals in Eq. 5 have a simple form based on exponent values and the distance between atom centers (Good et al., 1992).
Conformations of rings can also be examined in torsion space by including ring closure bonds in the search parameters. These bonds are broken during search calculations, allowing the torsions of the now pseudo acyclic structure to be freely adjusted. In order to properly close the ring, closure constraints are applied. These constraints maintain the distance between the two atoms forming the ring closure bond and the bond valence angles, within reasonable tolerances.
Systematic Conformational Searching on Ring Systems
|
During search calculations, when the ring closure bond is temporarily cleaved, the chirality of the closure bond atoms can be lost if these atoms are stereo centers. To account for this, and also to properly reorient in space the side chains off the 2 closure bond atoms, a torsional rotation is applied to realign the original closure bond axis with the new closure bond axis position. This torsion is called an implicit torsion, and is automatically handled by the search algorithm (Lipton and Still, 1988).
Since there can be large geometry variations localized at the two closure bond atoms, it is strongly recommended that you optimize the geometries of the conformers so that these deviations are distributed over the entire structure.
When searching conformations of ring systems, you must specify which ring closure bonds will be broken during search calculations. This is done using the SC_Search/ Set_Ring_Closure command. As explained on page 2, two implicit torsions are associated with each closure bond. These torsions are automatically assigned by the search algorithm.Therefore, they are not allowed as rotatable bonds.
Note that a ring closure bond must be in a unique ring, and if the bond is broken, its two adjacent ring bonds (implicit torsions) can rotate freely without breaking other rings. A fusion bond (a bond whose atoms are in two or more rings that share two or more atoms) cannot be defined as a ring closure bond. In addition, bonds adjacent to fusion bonds or connecting spiro atoms in spiro rings are not allowed as ring closure bonds.
Energy calculation or minimization is performed on a conformer using Discover functionality, after the conformer has passed checking for van der Waals clashes and satisfied any distance constraints.
Use of Energy Criteria in Systematic Conformational Searching
If energy evaluation is requested for a search, the final number of conformers found is restricted by the user-defined maximum number of conformers and energy threshold. The energy threshold is used first, to screen out the conformers whose energy values are a specified amount above the minimum energy--the energy value of the most stable conformer among all the conformers. If the number of conformers passing this phase of screening is too large, only the most stable conformers are selected, with the number of conformers being limited to the user-specified maximum value.
When energy minimization is performed, the resulting conformers are sorted according to their energy values from lowest to highest, and the atomic coordinates are stored in an archive (.arc) file.
If desired, duplicate conformers, whose energy and coordinates are nearly identical to those of an already-stored conformer, can be screened out. You can control this screening process by specifying the thresholds for the energy and rms values.
Note that there is no constraint on the torsion angles and distances during the minimization process. Therefore, the minimized conformers may not satisfy the original distance constraints, if any were present.
Discover functionality is used to calculate and minimize the energy. Several minimizers are used in sequence: steepest descents, followed by conjugate gradients, followed by a quasi-Newton-Raphson method known as BFGS (formerly called VA09A). Maximum derivative criteria are used to control when to switch to the next algorithm in the cycle: conjugate gradients is started when the maximum derivative falls below 10, and BFGS is started when the maximum derivative falls below 1.0. The total number of iterations for the complete minimization cycle and the final convergence criterion--the maximum derivative--are specified by the user. Note that optimization terminates when the total number of iterations is exceeded even if the convergence criterion has not been satisfied. It is also possible to specify whether or not to include charges or cross terms in the Discover energy calculation. Please refer to the Discover documentation for additional details.
Due to the nature of systematic searches, it is likely that atoms will occupy the same position in space in more than one valid conformation resulting from the search. Consequently, the same line segment in a vector map may occur in multiple trajectory frames loaded from a search. When this occurs, a duplicate vector is not generated, but instead a list of associated frames is generated for each vector of the vector map. You may use the Conformer/Display command to view any or all of the associated frames of a given vector.
Vector Mapping
Much of the computational work within Accelrys products is performed by background jobs that are run using the Insight II program as the user interface. Background jobs run concurrently with the interactive Insight II program; this is possible because, once started, they do not require user interaction. If you have access to more than one computer (mainframe or workstation), you may want to run some of the background jobs on a different computer (the remote host) than the one that is running the Insight II software (the local host).
Background Jobs
Running the background job on a remote host involves several general requirements:
The issue of file compatibility is primarily problematic when the data are not represented as text (ASCII). These files are called binary files and typically store data in a machine-dependent way, which varies from one host type to another (e.g., IRIS and Convex use different formats for some types of data). This problem is most often addressed either by using an additional program that filters the data or by converting the actual file formats to use machine-independent representations--especially those specified by Sun called XDR (external data representation).