jump to navigation

CoMFA – MFer. March 8, 2007

Posted by thwalls in Blogroll.


 So I’m finishing up some of my work and of course a CoMFA is required since my work centers around the inhibition of a particular transmembrane protein.  Of course, if you’re familiar with medicinal chemistry, you’ll know that every medchem student is entrenched with the thorough understanding that if the computer tells you the analog is good, then obviously…it’s good.  This then leads to your PI telling you to make all 150,000 analogs that could possibly be good since, hey, the in vitro data may be different.

I hate Sybyl, and I think it equally hates me. 

And possibly you too.



1. Mitch - March 8, 2007

What is CoMFA?


2. Egon Willighagen - March 8, 2007


CoMFA: Comparitative Molecular Field Analysis. A method that described the interaction of the molecule (ligand) with the protein (target), by making a grid of points for which an interaction field value is being calculated. Partial Least Squares (PLS, normally) is then used to make a regression model of all those grids points for many molecules against experimentally measured binding affinities. PLS returns weights for all grid points, where high weights indicate an important point. That leads to blue and red areas in the figures (I think, because not actually worked with Sybyl myself 🙂

3. thwalls - March 8, 2007

Comformational Molecular Field Analysis. Essentially, you make a slew of compounds based off of a given lead molecule. Then you test all of these different molecules in vitro against your target protein and you get IC50 values for all of the modified analogs. You can then take these modifications and put them into SYBYL which is a program that runs CoMFA along with the inhibition data. From this the computer can give you a model that displays the fields in the binding pocket that you should explore further with positively/negatively charged groups, hydrophobic bulky groups, or portions of the binding pocket that should be avoided all together. Essentially, it shows you your Qualitative Structure-Activity Relationship data in 3D (aka 3D QSAR). So from this you can say, “I need to put a methoxy group approximately 5 angstroms from that aromatic ring so I can hit that position in the binding pocket.”

So on the above picture, the blue and red blobs are portions of the binding pocket that are involved in binding with the molecule shown.

4. Wavefunction - March 15, 2007

“if the computer tells you the analog is good, then obviously…it’s good”
On the contrary…if the computer tells you the analog is good, you should be more skeptical and scrutinise it more closely!
(For the record, I am a modeler 🙂
Sybyl can be very difficult sometimes I agree. The thing for which I like it the most is for constructing those fabulous lipophilic Connolly surfaces. It’s worth 2300$ a year I think.

5. Great Molecular Crapshoot - June 27, 2007

CoMFA can be quite sensitive to molecular alignment. This becomes more of an issue if you are performing the analysis on multiple chemotypes. Ionisable groups are often modelled as neutrals because formal charges can easily swamp gas phase electrostatics. However if you’ve got mixed neutral and charged species this expedient may not be an option. Beware of PLS because cross validation is no guarantee that you will not overfit. The last journal submission that I bounced featured a CoMFA that had residuals for pIC50 that were typically less than 0.03. Definitely less than the uncertainties associated with the original enzyme inhibition results. However lots of BS still gets thru.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s

%d bloggers like this: