QSAR

5       Working with the Study Table

QSAR+ uses a study table to maintain and display the data for your QSAR analyses. Using the study table, you enter biological activity data, calculate descriptors for molecular structures, select data for graphing and statistical operations, and generate QSAR equations.

This chapter describes

The unique features of the QSAR+ study table are covered in the following topics:

Overview of the study table page 93

Study table menubar page 96

Using study table shortcuts page 104

Basic study table operations page 105

Overview of the study table

Just as when you work with other Cerius2 tables, you use scroll bars, cursor keys, and the mouse to navigate through a Study Table, and use the edit window at the top of the study table to enter and edit numeric and textual values. Similarly, you use the table tool bar to perform activities such as inserting and deleting data; cutting, copying, and pasting data; working with named groups; finding and sorting data; creating graphs; and changing the format and appearance of table cells, rows, and columns. You can also use the mathematical function library and formula capabilities available in all Cerius2 tables to transform and create table columns. For detailed information about working with Cerius2 tables, refer to the Cerius² Modeling Environment.

A distinctive feature of the QSAR+ study table is the presence of a menubar at the top. This menubar contains pulldowns with commands to execute all the QSAR tasks, including opening and saving files, editing the table, adding molecules and descriptors, defining dependent and independent variables, using graphical, statistical tools, and setting preferences. A specially customized tool bar allows quick access to the more frequently used commands. The components of a study table are described in the following sections.

Study table components

Figure 4 . Study Table

Menubar

The Study Table menubar contains specialized pulldowns to access all the QSAR+ functions. Pulldowns included are: File, Edit, Molecules, Descriptors, Variables, Tools, and Preferences. Notice that the same pulldowns, with the exception of the File pulldown, which is replaced by Show Study Table, appear as menus on the QSAR card.

Tool bar

The QSAR tool bar is unique to the study table. Some of its tools are specific to QSAR+ and some are present in all Cerius² tables. You can use these tools to quickly access QSAR+ functionality.

Methods popup

Use the Methods popup to specify which of nine different statistical methods you want to use in order to generate a QSAR equation or to explore the data. The selected method is then used by QSAR+ when you click the RUN button. Briefly, your choices are:

• GFA-- Genetic function approximation
• G/PLS -- Genetic/partial least squares
• Linear -- Multiple linear regression
• PLS -- Partial least squares
• Simple -- Simple linear regression
• Stepwise -- Stepwise multiple linear regression
• PCA -- Principal components analysis
• PCR -- Principal components regression
• Cluster-- Cluster analysis

Indicates the current defaults set for the QSAR study table. The default sets define general preferences regarding molecules, descriptors and statistical methods. Three sets are predefined, corresponding to QSAR, COMBICHEM, and QSPR applications.

Edit window

You use the edit window to enter and modify the values in numeric and textual data cells. If a cell value can be edited, it is displayed in the edit window when the cell becomes current. For detailed information about the use of the edit window on Cerius2 tables, refer to the Cerius² Modeling Environment.

Variable indicators

Variable indicators appear at the top of each study table column that contains a dependent (Y) or independent (X) variable. As you select and clear variables, these indicators make it easy for you to determine which Study Table columns contain the dependent and independent variables used in your QSAR analysis.

Row

Each row in the table represents an observation (or experiment), and consists of a molecular structure, one or more measures of biological activity, and one or more calculated descriptors. Study Table rows can also contain data that you enter and conformational data (such as the lowest energy conformation).

Column

The first study table column contains molecular structures that can be represented as pictographs. Other study table columns contain measures of biological activity, the results of descriptor calculations, and the results of QSAR equation generation (that is, values for predicted activity and residuals). You can create and use additional columns to enter and explore data from experiments and other sources.

Study table menubar

File pulldown

• New. Deletes all data from the current study table and creates a new one with the three predefined columns, Structure, Activity, and Structure Name.
• Open... Deletes all data from the current study table and opens a file browser to select and open a table previously saved to disk.
• Save. Saves the current study table as a table file, using a previously defined filename. If the filename has not been defined, the Save As... menu item is used instead (see below). The defined name is used in subsequent study table Save operations.
• Save As... Brings up a file browser to define a name for the study table and to save it as a table file.
• Import... Brings up the generic Cerius² Import Table control panel to import data from an ASCII file into the study table.
• Export... Brings up the generic Cerius² Export Table control panel to export data in the study table to an ASCII file.
• Reset. Removes all data from the study table and resets all options to the default values.

• Cut. Generic Cerius² Tables command. Cuts the selected rows/columns, saving them in the clipboard and deleting them from the study table.
• Copy. Generic Cerius² Tables command. Copies the selected rows/columns to the clipboard.
• Paste. Generic Cerius² Tables command. Pastes the data from the clipboard to the selected rows/columns.
• Paste Special... Brings up a panel to paste data from the Table Manager into the study table. Options are provided to match rows/columns when pasting the data.
• Clear. Generic Cerius² Tables command. Clears the values of all selected cells.
• Insert. Generic Cerius² Tables command. Inserts rows (if rows are selected) or columns (if columns are selected) after the current cell position. If no rows or columns are selected, you are asked about the insertion.
• Delete. Generic Cerius² Tables command. Deletes the selected rows/columns from the study table without saving them to the clipboard.

Molecules pulldown

• Add All. Adds all the molecules in the Cerius² Model Window to the study table. This can also be done by clicking the corresponding icon in the study table toolbar.
• Add Selected. Adds all the selected molecules in the Cerius² Models window to the study table.
• Add Current. Adds the current molecule in the Cerius² Models window to the study table.
• From SD File.... Brings up a panel that allows you to add molecules from an MDL SD file to the study table. Data fields contained in the SD file can optionally be added as columns to the study table. You can also choose whether to add all the molecules in the file, a range of molecules, or just a single molecule.
• From SMILES File... Brings up a panel that allows you to add molecules from a SMILES file to the study table. You can choose whether to add all the molecules in the file, a range of molecules, or just a single molecule.
• Export to SD File.... Brings up a panel that allows you to export data from the study table to an SD file. Both molecules and numerical or text data can be exported.
• Get Models by Name Adds models from the Cerius² Model Manager to study table rows that have no valid models assigned to them. The name in the Structure Name column is matched with the name of the model in the Model Manager.
• Recover Molecules.... Brings up a panel with functionality to use information stored in the study table to recover molecules that have been deleted from the Cerius² Models window and from the study table. This information can be either the file name, file type and file index of the molecules (if originally read in from a file), or core structure and R-groups (if originally built by the Cerius² Analog Builder).
• Core SSS.... Brings up a panel to search for a core structure in the molecules currently added to the study table. This allows the identification of "fragments"' in the molecule: parts of the molecule not in the core, attached at different R-group positions. For more on fragments and core structure searches, see Chapter 8, Working with Fragment Constants.

• Add Default. Adds all the default molecular descriptors from the current descriptors database to the study table. This command can be accessed using the corresponding icon in the study table toolbar.
• Select.... Brings up a panel to select molecular descriptors from the different classes or families to the study table.
• Databases.... Brings up a panel to display and edit the molecular descriptors databases, allowing the user to change the default status of specific descriptors, and to add new descriptors.
• Fingerprints... Opens a submenu that gives access to control panels that provide a faster way of calculating MDL ISIS keys from an SD file or Daylight fingerprint from a SMILES file and loading the results into the study table.
• FragConst Selection... Brings up a panel to select fragment constants (Hansch-type classical substituent constants, including sigma-para, sigma-meta, sterimol parameters and others) as molecular descriptors and to add them to the study table.
• FragConst DB Editor... Brings up a panel to display and edit fragment constants databases, allowing you to include new fragments and their associated constants, to edit existing constants or add new user-defined ones, and to save the data to a new database.
• Recalc Descriptors. Recalculates the molecular descriptors for the selected columns/rows. If no columns or rows are selected, the descriptors corresponding to the independent variables are recalculated for all rows.

• Set X. Sets the selected columns in the study table as independent variables (X).
• Set Y. Sets the selected columns in the study table as dependent variables (Y).
• Clear X. Removes the selected columns in the study table from the list of independent variables (X). If no column is selected, all columns are removed from the independent variables list.
• Clear Y. Removes the selected columns in the study table from the list of dependent variables (Y). If no column is selected, all columns are removed from the dependent variables list.
• Clear XY. Removes the selected columns in the study table from the list of dependent and independent variables. If no column is selected, all columns are removed from the dependent and independent variables lists.

• Manage Variables... Brings up a panel to facilitate the visualization and selection of independent and dependent variables.
• Manage Independent... Brings up a panel to facilitate the selection of independent variables, including options to select those columns with the highest variance or with a highest correlation with activity. This panel also has controls to display independent variables corresponding to 3D-QSAR descriptors (such as field values from molecular field analysis (MFA) or interaction energies from receptor surface analysis (RSA) in the Cerius² Models window.
• Manage Observations... Brings up a panel to select which rows in the study table to use as observations (samples) when running a QSAR statistical analysis or regression method. The default is to use all rows in the study table.
• List Status. Prints information to the text window regarding the columns and rows in the study table currently set as dependent variables, independent variables, and observations. For more information, see Chapter 11, Working with Variables and Observations.

• Properties.... Brings up a generic Cerius² Tables control panel to set general table properties, such as formats, data types, headings, alignments, etc.
• Groups... Brings up a generic Cerius² Tables control panel to create, manipulate, and display study table groups, which are sets of rows/columns you define.
• Find... Brings up a generic Cerius² Tables control panel to search for and optionally replace study table data.
• Sort... Brings up a generic Cerius² Tables control panel to sort study table rows based on the values of one or more selected columns.
• Restore View. A generic Cerius² Tables function to restore the display of the study table so that all the rows and columns are displayed.
• Recalculate. A generic Cerius² Tables function to recalculate all columns that contains derivations for all rows in the study table.

• Histogram Plots. Generates a histogram plot for each of the selected columns in the study table. If no column is selected, histograms for all columns marked as independent variables are generated.
• RunePlots. Generates a rune (distribution) plot for each of the selected columns in the study table. If no column is selected, rune plots for all columns marked as independent variables are generated.
• 2D Plot.... Brings up a generic Cerius² Tables control panel to create 2D plots from data in the study table.
• 3D Plot... Brings up a control panel to generate a 3D plot of specific columns and rows in the study table. The 3D plot is displayed in the Cerius² Models window.

Tools/Statistical menu item

• Correlation Matrix. Generates and displays a table with the correlation matrix for the molecular descriptors in the study table. The correlation matrix is generated for all columns marked as independent or dependent variables. If there are no dependent or independent variables, the correlation matrix is calculated for all numeric columns.
• Summary Statistics. Creates a table with descriptive statistics for columns in the study table, including mean and median, variance and standard deviation, minimum, maximum, and range, sum, sum of squares, kurtosis, skewness, and count. The descriptive statistics are generated for all columns marked as independent or dependent variables. If there are no dependent or independent variables, the descriptive statistics are calculated for all numeric columns.
• Validate QSAR... Brings up a control panel with tools to validate a QSAR equation, including cross-validation and randomization tests.
• Find Outliers... Brings up a control panel to allow you to identify and manage outliers in the QSAR equation. There are options to remove the outliers from the observation group and refit the QSAR equation.

These actions operate on entire columns:

• List Empty Cells in the selected columns or, if no column is selected, in the columns marked as independent or dependent variables (XY columns).
• Fill Empty Cells in the selected or XY columns with a value. The value can be:

  The mean value for the column.
  A user-specified value.
  The value in the corresponding row of a specified column.

• Remove Columns with empty cells from the XY columns groups.
• Remove Rows with empty cells from the observations group.

Other Tools menu items

• Equation Viewer.... Brings up the Equation Viewer control panel to display and manipulate the QSAR equations generated in the current Cerius² session.
• Conformers... Brings up a control panel to display information about conformers for molecules in the study table.

• Defaults Set. Resets default values for options affecting the behavior of operations on molecules, descriptors, and statistical methods. Three predefined defaults sets are provided: QSAR, COMBICHEM, and QSPR. Other user-defined defaults sets can be specified.
• General... Brings up a control panel to set values for general QSAR options, such as whether to display the study table toolbar, whether larger numbers mean higher activity, and whether to automatically validate QSAR equations and display tables and graphs for ANOVA and beta-coefficients results.
• Molecules... Brings up a control panel to set values for options regarding molecules, such as whether to add hydrogens, minimize the energy, calculate charges, and generate conformations when adding molecules to the study table, and whether to immediately recalculate descriptors when molecules already added to the study table are edited.
• Histograms... Opens a control panel to set preferences for histograms. For example, you can allow Cerius² to define the bins automatically or set your binning preferences manually. (Histograms can be made from a BDF file, if one is selected, or from selected rows, rows with observations, or all rows in the study table.)
• Statistical Methods... Brings up a panel to set values for options affecting the different statistical methods available in QSAR+.

The QSAR study table also has some unique shortcuts associated with it:

Action	Result
Double-click a structure	The structure is brought into the Cerius2 Models window and becomes current.
Double-click a column containing conformer information	The Conformers table is displayed. The Conformers table contains information for the structure associated with the row that you double-clicked.
Double-click a column label	Descriptive information for the table column is displayed. This information includes the number of and label assigned to the column, a description of the column contents, and, where applicable, the formula associated with the column.

Basic study table operations

• Displaying the current study table.
• Saving your work.
• Accessing a new, blank study table.
• Reloading an existing study table.
• Opening other table files.
• Exporting a study table.

• Display the current study table if it is hidden.
• Display a blank study table if you have not yet created or loaded a study table in the current Cerius2 session.

Do either of the following:

• Choose Show Study Table from the QSAR menu card.

or:

• Choose Show Study Table from the MFA, MSA, or DRUG DISCOVERY card.

Saving your work

To save a Cerius2 session

1.   Choose the File/Save Session menu item on the Cerius2·Visualizer menu bar.

The Save Session control panel appears:

2.   Use the file-selection tools to specify the location and name under which you want to save the current Cerius2 session.

For detailed information about using the Save Session control panel, refer to the Cerius² Modeling Environment.

3.   Click Save.

The entire Cerius2 session (including the study table, its data, and its molecular structures) is saved under the name you specified: Cerius² writes a series of .msi files in the current directory. If files of the same name already exist in this directory, they are overwritten.

To access a new, blank study table

Determine whether a study table exists. Then:

• If a study table does not yet exist (that is, if you have not yet created or loaded a study table), follow the instructions provided in Displaying the current study table
• If a study table already exists, you can do any of the following: -o- -Chooseth File/New menu item in the study table menubar to clear the existing study table.

  -o- Choose the File/Reset menu item in the study table to clear the existing study table and reset all options to their default values.

  -o- -Choose the File/New Session menu item in the Cerius2·Visualizer to reinitialize Cerius2. For information about Cerius2 sessions, refer to the Cerius² Modeling Environment.

Warning

Warning

1.   Choose the File/Load Session menu item in the Cerius2·Visualizer.

The Load Session control panel appears.

2.   Use the file-selection tools to identify the Cerius2 session containing the study table that you want to reload.

For detailed information about using the Load Session control panel, refer to the Cerius² Modeling Environment.

3.   Click Load on the Load Session control panel.

The entire Cerius2 session is reloaded. You can now access the previously created study table (for example, by choosing Show Study Table from the BUILD QSAR card).

• If you have study tables created using the QSAR+ application, and you want to work again with those study tables.
• If you have a scientific paper that contains a numeric dataset and you have typed that data into a table. You now want to use the genetic function approximation algorithm (described in Chapter 13, Genetic Function Approximation) to analyze the data in that table.

To open a table file

1.   Choose the File/Open... menu item in the study table menubar.

The Open Study Table control panel appears.

2.   Use the file-selection tools to identify the table (.tbl) file that you want to open as a study table.

For detailed information about using file-selection tools, refer to the Cerius² Modeling Environment.

3.   Click Open.

QSAR+ displays the specified table as a study table. All cells in column 1 (Structure) of this study table are blank.

To export a study table

1.   Choose the File/Export... menu item in the study table.

The Export a Table control panel appears.:

Figure 5 . Export a Table control panel

2.   Use the file-selection tools to identify the file that you want to export.

3.   Customize the export by:

-o- Optionally customizing the export format.

-o- Indicating whether you want to export all or a portion of the specified study table.

The procedure for exporting a study table is identical to that for exporting any Cerius2 table. For detailed information about using the export options shown on this control panel, refer to the Cerius² Modeling Environment.

4.   Click Export.