Build Field-Based Model Dialog Box

Build Field-Based Model Dialog Box Features

 

• Field style option menu
• Use input partial charges option
• Maximum PLS factors text box
• Grid spacing text box
• Extend grid by N Å beyond training set limits text box
• Ignore force fields within N Å of any training set atom option and text box
• Truncate steric force fields at text box
• Truncate electrostatic force fields at text box
• Eliminate variables with StdDev < text box
• Eliminate variables with |t-value| < option and text box
• Number of ligands to leave out for cross-validation text box

Field style option menu

Select the fields that you want to include in the model.

• Force field—Use the force-field electrostatic and steric fields for the model (CoMFA).

• Gaussian—Use the five standard Gaussian fields for the model, excluding aromatic ring fields (CoMSIA).

• Extended Gaussian—Use all the Gaussian fields for the model, including aromatic ring fields.

• Custom—Select the combination of force-field and Gaussian fields to use for the model. To make the selection, click Edit, and select the fields from the list in the Custom Field Style dialog box. In this dialog box you can also import feature definitions for the fields.

Use input partial charges option

Use the partial charges stored with the input structures for the electrostatic fields, rather than the partial charges from the OPLS_2005 force field.

Maximum PLS factors text box

Specify the maximum number of partial least squares factors to use in the model. A model is built for each number of PLS factors up to the specified maximum. You can choose any value, but you should not generally set this value to more than N/5, where N is the number of ligands in the training set, because of the risk of overfitting. To avoid overfitting, the standard deviation of regression should be larger than the experimental uncertainty in the activity values for the model that you use.

Grid spacing text box

Set the spacing of the grid points on which the fields are evaluated, in angstroms.

Extend grid by N Å beyond training set limits text box

Extend the grid for the QSAR model so that the outermost points of the grid are at least the specified distance from any atom in the training set. This allows you to make predictions for ligands that are larger than the training set ligands.

Ignore force fields within N Å of any training set atom option and text box

Set the distance from the training set atoms inside which grid points are discarded when evaluating the fields. A grid point is discarded if it is within the specified distance of any of the atoms of any of the training set molecules. This option essentially prevents the model from being dominated by the strong fields close to the nuclei.

Truncate steric force fields at text box

Specify the energy cutoff value for truncating steric force fields, in kcal/mol. Any field that is greater than this value is set to this value.

Truncate electrostatic force fields at text box

Specify the energy cutoff value for truncating electrostatic force fields, in kcal/mol. Any field that is greater than this value is set to this value.

Eliminate variables with StdDev < text box

Eliminate variables whose standard deviation is less than the value given in the text box. The smallest value you can set in this text box is 0.001. This allows you to eliminate variations that are not significant.

Eliminate variables with |t-value| option and text box

Select this option to use a t-value filter to eliminate independent variables whose regression coefficients are overly sensitive to small changes in the training set composition, and enter the threshold for eliminating variables in the text box. The resulting models have fewer uninformative variables and tend to give better predictions on test set compounds.

Number of ligands to leave out for cross-validation text box

Set the number of ligands to be used in the leave-N-out cross-validation statistics. This value is also used for assessing the stability of the model (Stability property). The default is 1.