
QSAR COMFA
CoMFA Analysis
Function:
COMFA is a SYBYL/QSAR method involving the shapes of molecules. The COMFA command invokes operations involving alignments, fields and regions, which are entities specific to CoMFA studies.
Menubar:
MSS: QSAR >>> Manage CoMFA...
Format:
QSAR COMFA class [additional_input]
Arguments:
Definitions:
- CoMFA = (Comparative Molecular Field Analysis) operates on energy values at points in space surrounding the molecules of interest.
- REGION = a set of boxes (from a regular grid or lattice down to a single point) with specified probe charge and atom type.
- FIELD = set of values, one value (energies, PLS coefficients, or CoMFA summary STDEV*COEFFICIENTS) for each point in the region.
- A CoMFA analysis is successful only if molecules are aligned properly within the region.
Steps:
CoMFA (Comparative Molecular Field Analysis) involves four steps:
This technique may appear overwhelming at first glance. However, it is actually straightforward once you have seen it applied. To that effect we recommend that you start with the CoMFA demonstration script (in command mode, type DEMO COMFA).
A key aspect in a CoMFA study is the aligning of molecules. AUTOCOMFA is usually a good starting point for any CoMFA analysis.
- Prepare a database containing the molecules in the study.
- Align the molecules so the important portions of the molecules line up similarly when overlaid. Your judgment may be augmented by
- the FIELD_FIT command (COMFA ALIGNMENT DEFINE FIELD_FIT) and by least squares rigid body fit (top level command FIT). While you can use molecules as stored in the database, it is safer to create named alignments to allow multiple alignments of your molecules.
- Use the COMFA REGION DEFINE command to prepare a region, the set of points in space where energy probes are desired, with additional parameters such as atom type and charge of the probe.
- Create and save energy fields (optionally). This step is required to graph the fields or to average conformations into a composite field.
- Create a table (TABLE CREATE name DATABASE) with the molecules to be studied. Add data columns for the dependent variable (e.g. activity) and for CoMFA fields, as well as information such as logP if desired.
- Fill in the table, via TABLE ENTER and TABLE EVALUATE.
- Perform a PLS analysis (ANALYSIS DO... PLS) on the columns. The first run is normally with crossvalidation to find the optimal number of components before the final run without crossvalidation.
- Iterate to improve the model (optionally), by adjusting the molecule's alignments, by dropping or adding molecules, etc.
- Examine the results graphically (COMFA FIELD RETRIEVE, COMFA FIELD GRAPH).
- Predict the activity of new compounds (ANALYSIS PREDICT).
Additional Information:


Copyright © 1999, Tripos Inc. All rights
reserved.