QSAR - Working with Fragment Constants

QSAR

8 Working with Fragment Constants

Hansch-type QSAR studies require extensive work with molecular topologies (2D structural information). This is because constants are assigned to different substituent positions based on the fragments present at those positions. Thus, each molecule in a study must be partitioned into a (common) core and numbered substituents. There is also a need to associate different constants with different fragments. The substituent numbering can be complicated by topological symmetry, which can result in more than one way to assign constants.

Functionality is available from the Cerius² study table to:

Find a specified core within each molecule and label each substituent
Select one or more databases for lookup
Select constant types for use on a per-substituent position basis from a list of available constants for the currently selected set of databases
Use unique SMILES to look up constants in a database based on topological structure of the fragments whenever descriptors are calculated
Generate fragments from a study not found in the default database so new constants can be entered into a custom database that you are building
View and edit databases, as well as calculate sterimol parameters for new fragments
Generate statistics regarding substituent positions to suggest alternative core definitions

Before proceeding into this chapter, you should work through the tutorial lesson on fragment constants in the Cerius² Tutorials-- Life Science book.

This chapter discusses:

There are three main tasks in working with fragment constants; each one is associated with a control panel:

You can access these control panels using the menu bar on the Study Table.

Specifying R-groups

The substituent positions defined for fragment constants are referred to as R-groups, which is consistent with usage in the Analog Builder. Before doing a fragment constant analysis you must have a core model and it must not be entered into the Study Table. You identify the R-group positions on the core model using the Analog Builder's R-group tool, available from the BUILDERS 2 card deck.

Selecting fragment constants

There are any number of different constant types available: you need to specify which ones are to be used at which positions. You also may wish to change or add databases to the search list. You accomplish these tasks with the Fragment Constants Selection control panel.

Accessing the Fragment Constants control panel

To access the Fragment Constants Selection control panel, open the study table (available from the QSAR card deck) and select Descriptors/FragConst Selection... from the menu bar. When the Fragment Constants Selection control panel opens, its top list box contains a list of databases. Highlighting indicates those databases that will be used for searches. The constant types available in those databases are shown in the lower list box, which is dynamically updated according to what is selected in the database list.

Adding a database

If you select ADD DATABASE... a control panel appears so you can append a database to the end of the database list. The last database listed is searched first, followed by next-to-last, etc. If you want to search in a different order, you can delete the higher databases and then re-add them so they appear at the bottom of the list (and are therefore searched earlier). Multiple entries for the same fragment are allowed in different selected databases or within the same database. The last entry found in the last database listed has priority during lookup.

Specifying the core model

Since different constants can be assigned to different R-group positions, you must register the core model with this control panel. Do this using the Set Current Model as Core button. If you add additional R-groups to the core after registration, you must re-register the core. Then you can assign constants to the new positions. The left- and right-pointing triangles are used to change the current R-group for assigning constants. You can control whether highlighting a constant specifies it for all positions or just the current one with the Select Constants for popup.

Adding constants to the study table

After you select all the desired constants, you can add them to the study table with the Add Selected Fragment Constants to Study Table button. This triggers their lookup in the database(s) for molecules already present in the study table. If the molecules in the study table do not yet have R-group positions specified and named, you should first complete that task, using the Core Substructure Search control panel. Once the R-group positions are defined for the study table molecules, you can proceed with adding constants to the study table.

Identifying fragment positions in study molecules

To know what fragments to search in the database and to which R-group positions the retrieved constants should be applied, you need to assure that a mapping is specified between each study molecule and the core. The mapping identifies the location of each substituent, as well as identifying all of the atoms that correspond to the core.

Accessing the Core Substructure Search control panel

To open the Core Substructure Search panel, select Molecules/Core SSS... from the study table menu bar.

How you use the Core Substructure Search control panel depends on the origin of your study molecules:

If your study molecules were created using the Analog Builder and the default fragment constants fragment library, you do not need to use this control panel at all.
If you used the Analog Builder but with a different fragment library than the one you want to search, use the Core Substructure Search control panel to name the fragments in your molecules, see Renaming fragments (next section).
If your molecules come from a source that does not indicate which atoms are core and which substituent, see Core searching.

Renaming fragments

If you used the Analog Builder but with a different fragment library than you want to search, use the Core Substructure Search control panel to name the fragments in your molecules so that they correspond to the fragment names in the fragment constant database(s). The Name Fragments by Database Lookup button performs a unique SMILES-based lookup of the substituent at each R-group position, finds the corresponding database names, and adds these to the models as well as posting them to the study table columns R1, R2, etc. If you need to manually override this mechanism (currently you must do this to make stereoisomers unambiguous), edit the contents of these columns and then click Name Fragments by Study Table Columns.

Core searching

If your molecules come from a source that does not indicate which atoms are core and which substituent, you need the core search facilities.

To perform a core search, sketch a core model with hydrogens for substituents. Specify R-group positions by marking the appropriate hydrogens using the R-group specification tool available in the Analog Builder control panel. Then register the core model using the Set Current Model as Core button on the Core Substructure Search control panel. This action also communicates the core model choice to the Fragment Constants Selection control panel.

Clicking the CORE SEARCH button causes a search for the core structure in each molecule in the study table (or selected molecules, if some rows are selected). Since the fragment constant approach is a 2D analysis, the matching is done purely by connectivity, ignoring 3D coordinates. If any molecule does not contain the core or if there are multiple topologically distinct ways to perform the match, a message appears in the text window. For instance, only a single match of biphenyl is reported, as a benzene core with one R-group. All topologically equivalent matches are automatically filtered out. If the Orient to Core box is checked, the molecules are also oriented so that their core portions align with the core model.

The results of the search are summarized in the Core SSS column in the study table. The text M of N in a cell indicates that match M is currently selected out of N total matches. Double-clicking a cell in this column brings the molecule in the corresponding row into view in the Cerius² Models window, with the match displayed. The core atoms are highlighted and the fragments are labeled by R-group position.

If the Orient to Core box is checked, the molecule is also re-oriented so that the current core match aligns with the core model. Double-clicking the same cell again causes the next match choice to be selected. This is reflected in a new value of M in that cell's M of N match counter. If M was already equal to N, then M wraps around to 1.

Once you are happy with your match choices, click Name Fragments by Database Lookup to associate database names with each fragment. If any fragments are not present in the database, they are created and added to a special database overwriting the file tmp.UNKNOWN.fdb, located in the current directory. To view the fragments in this file, click View Unknown Fragments in Database Editor. The database editor panel appears with the tmp.UNKNOWN.fdb database loaded. The columns @R1, @R2, etc., are statistics regarding the frequency of occurrence of each fragment in the study molecules. The value in the cell is the number of times the fragment occurs at the given R-group position. This information often suggests a different definition of the core model and/or refinement of your study set. For an example, see the Fragment Constants tutorial lesson in the Cerius2 Tutorials-- Life Science.

Editing the fragment constants database

To make use of your own fragments and constant values, you need to add them to a database. You do this with the Fragment Constants Database Editor control panel.

Accessing the Fragment Constants Database Editor control panel

To open the Database Editor control panel, select Descriptors/FragConst DB Editor... from the study table menu bar. This opens an empty edit buffer in the form of a Cerius² table in which you can use all the standard table tools.

To edit a particular database, you must read it into the buffer. First, open the file-browsing control panel by clicking Open FragConst DB.... and selecting a filename. Database files have the extension .fdb. They are designed to be difficult to corrupt, by just being a concatenation of entries. New entries are simply appended to the file, which is in ASCII format. There is also an index file associated with each database that has the extension .fdbind. It is automatically updated every time the database file is written out and can, in general, be ignored. (If the index is corrupted because of a disk crash occurring in the middle of an update it can easily be regenerated by reading the .fdb file into the buffer and then immediately writing it back out.)

You can read the default database into the buffer by selecting the file Cerius2-Resources/QSAR/FragConst/hansch.fdb in the file browser and pressing OPEN. You can edit constant values or add new constant types by inserting new columns. These constant values and types can be conveniently cut and pasted from other Cerius² tables.

Use the ADD command on the Database Editor control panel to read in new fragment databases. If you need to specify the substitution point for the core model, use the R-group tool. Then, on the fragments, the appropriate hydrogen is replaced with an X atom. This can be done in the C²·Sketcher by clicking the periodic table icon next to the Edit Element button.

If you want to add new fragments that already occur in a set of study molecules with an associated core model, follow this procedure:

1. Deselect or remove all databases from the database list in the Fragment Constants Selection control panel.

2. Open the Core Search control panel and perform:

a. CORE SEARCH

b. Name Fragments by Database lookup; ignore the messages in the text window about missing fragments

c. View Unknown Fragments in Database Editor

Calculating sterimol parameters for fragments

Sterimol parameters can be calculated for each fragment in the edit buffer with the Calculate Sterimol Parameters button. It is important to optimize the geometry of the fragments before calculating these parameters, since they depend on the 3D structure. To make your changes accessible to the constant lookup procedure, they must be written out to a fragments database. Click Save FragConst DB... to open a control panel for this task. The Append to Database checkbox controls whether the selected database file is overwritten or appended to. When an append operation adds a duplicate entry for a fragment, the last entry is the one that is used on lookup.

If you want to delete entries you can read the database into the editor, delete unwanted rows, and save the database again with the Append to Database checkbox unchecked. The original database is silently overwritten, but a backup will be made first, with the extension .fdb.bkup.