4.2.1. CLASSES Menu
Generate
Allows you to generate classes using various automated options. The details of this menu item are discussed in the Generate Classes section.
Add Using Query
Equivalent to the Create Class Using Query option from the CHEMISTRY menu.
Scaffolds
Several of the submenu items are used to define class scaffolds. These include Define Using MCS, Define Using Markush MCS, Define Using Lasso and Define Using Query. These methods are particularly useful when no class scaffold is generated automatically, such as when classes are generated manually or after two or more classes are combined. They are useful with the Frameworks method, which by default can produce scaffolds smaller than the actual MCS.
The scaffolds can be searched using a user-defined query using the menu item CLASSES>>Scaffolds>>Search Using Query.
One or more scaffolds can be exported to a file using the menu option: CLASSES>>Scaffolds>>Export. The scaffolds can be saved in a number of different formats, including an SD file, SMILES file and CQF file.
R Tables
Provides options to Generate a new R table for a selected class or Delete an existing R table. There are also options to change the numbering of the R groups within an R table as well as to launch the R Group Analyzer window.
The Add Classes Using R Group Subsets option is used to create new classes that each contain a subset of the compounds in a single selected class. Specifically, each subset is chosen so that it simplifies the R table of the original class; it is therefore necessary to select a single class with an existing R table prior to using this feature. The approach to finding subsets is identical to the one described in the section entitled Finding Subsets Using R-Groups, and uses the same "Generate subsets" dialog displayed in that section. For each new class, a new R table is created automatically, and this R table will have fewer R groups than the original one, increasing the structural similarity of the class members and making it easier to perform subsequent SAR analysis. Note that this feature can also be activated via a button on the toolbar.
Subclasses
A subclass is a class whose members are all contained in another class called the parent class, and whose scaffold is larger (by atom count) than that of the parent class. Subclasses can be thought of as a “second layer” of classes, one level deeper than the original ones, and can be useful for further elucidating the structural motifs present in those original classes, particularly when the latter are large and contain significant structural diversity. Note that while the scaffold of a ADMET Predictor-generated subclass is required to be larger than that of its parent, the subclass scaffold need not be contained in the parent scaffold: the former is not necessarily “grown” from the latter.
This submenu provides options to hide and unhide subclasses, remove all subclasses, or move subclasses such that each one is directly below the class of which it is a subset.
Compare Classes
Provides a number of options to explore the relationship between classes. These are explained in a subsequent section.
Combine (Ctrl+Alt+C)
Allows you to combine two or more selected classes into a single superclass. If at least one of the selected classes has more than one member compound, you will be prompted whether such classes should be retained in the spreadsheet. Note that after combining classes, you should regenerate the scaffold of the superclass using one of the options under CLASSES>>Scaffolds.
Break Up (Ctrl+Alt+B)
Breaks up the selected classes into singletons: that is, removes the selected classes and replaces them with classes that each contain only a single molecule. Note that new singleton classes will not be created for molecules that are already contained in another non-singleton class that was not selected.
Replicate
Lets you create a new class by copying the contents of a selected class. The result is to create a new class having the same molecules and scaffold (if any) as the selected class. Note that no other information from the original class – such as R table – is copied to the new class. It is not possible to replicate a singleton class.
Align Compounds (Ctrl+Shift+L)
Aligns the molecules in selected classes so that they have the identical scaffold orientation. The final scaffold orientation of all molecules in a class will exactly match that of the single representative molecule for the class.
Prioritize
Assigns scores to the classes based on their scaffolds, property ranges and other attributes, then re-sorts the spreadsheet such that high scoring classes appear at the top. When selected, this menu item displays the Prioritize Classes dialog. First you must select a single numeric attribute (activity, for example) to use for scoring. You then need to specify a categorical weight for each of five pre-defined class properties: (1) large class size, where size is the number of member compounds, (2) high scaffold complexity, where complexity is defined as the sum of scaffold ring counts and bond counts, (3) large range of attribute values, (4) large maximal attribute value among the class members, and (5) large number of class members with a high attribute value. The right side of the dialog allows you to specify desired ranges for each of the five attributes. If a class falls outside of the range, it will not be assigned a score and will instead be sorted to the bottom of the spreadsheet once the process completes.
The final score for a class is based on comparing each of the five values described above to its distribution of values over all classes; values larger than the average contribute positively to the final score, while values less than the average contribute negatively.
Organize By Scaffold
This allows you to organize a set of classes into clusters, or “families.” This feature is particularly useful when the number of generated classes is too large to analyze by simple inspection. It organizes the classes by structural similarity of the class scaffolds, placing all classes with similar scaffolds into the same family and arranging them next to one another in the Classes tab spreadsheet.
Selection of this menu option opens a Generate class families dialog that provides a number of options to control the manner in which families are assigned.
-
The first option controls how similarity between classes will be calculated. The first three choices for this option correspond to different types of descriptors: “Linear paths” and "ECFP" indicate unbranched fragment and ECFP descriptors, respectively, derived using the data in the spreadsheet; “internal fingerprints” indicates a fixed set of descriptors that does not depend on the data set. The first two options tends to produce families with greater internal similarity, but may lead to a larger number of families and families of smaller size (i.e., fewer classes per family). Note that if a particular class has a scaffold, descriptors are derived using only that scaffold; otherwise they are derived using the entire Representative Structure for the class. The "class member lists" option indicates that class similarity should be measured by comparing the molecules contained in the classes: two classes sharing many common molecules would be considered similar, while those sharing few molecules would be considered dissimilar.
-
The next dialog option controls whether singleton classes are considered during the assignment of families. If this option is not selected, singleton classes are ignored, with the result that each singleton class is assigned a family tag (ID) of -1.
Once families are generated, you can display the family tag of each class by changing one of the Classes tab columns to Class Family.
Note that the Frameworks method of class generation performs scaffold organization automatically, so this option need only be considered for other methods of class generation.
Display Selection In Compounds Tab
Allows you to transfer the compounds within selected classes to the Compounds Tab. Once the transfer has occurred, the Compounds tab will be active and all compounds not contained in any of the selected classes will be hidden automatically. This operation can also be performed using the relevant toolbar button, as shown below.
Please keep in mind that this toolbar button is visible only after you have activated the Classes tab.
Change Representative Structures
By default, the Representative Structure column of the Classes tab displays the class member having the lowest molecular weight. This menu option allows you to choose a different criterion. You can choose a single molecular property and set each class representative to be the molecule with either the minimum or maximum value of that property.
Change Column Functions
Once classes are generated, the spreadsheet in the Classes tab shows for each class a single value for each attribute column even if the class contains more than one molecule. For numeric columns, this value is by default the average among all compounds in the class. This menu option allows the function (also called an aggregation) to be changed to, for example, the minimum, maximum, range, etc. Details of the available aggregation functions is listed elsewhere.
Change Chart Functions
Allows you to determine what is displayed in the Classes tab charts. The default option is "Show All Compounds," which causes all of the compounds to be displayed individually. If this is changed to "Average," for example, the charts will display classes rather than compounds, and the plotted values will be averages over class members.
Add Comments
Users can assign arbitrary text to one or more classes in the spreadsheet. This provides a useful means of labeling classes with important notes for future reference. Note that this option requires that at least one class be selected.
Add Attributes
Users can add any number of text attributes to classes in the spreadsheet. Class attributes are essentially identical to class comments, but can be assigned arbitrary names (for example, “John’s comments”). As with compound attributes, class attributes can be displayed in the Classes-tab spreadsheet, exported to file and searched using queries.
Modify Attributes
Allows you to modify or delete the class attributes. This option is similar to the one used for modifying compound attributes.