Combinatorial chemistry is not so much a field as it is a methodology for conducting scientific research. The key element of these types of techniques is developing a large library of different combinations of molecular parts. After that, you’ll need to develop a set of screening criteria and then test your library against those criteria. In some ways, getting a true positive “hit” in a library requires a bit of luck. This is partially why these techniques are sometimes looked down upon by people in fields such as structural biology. I, however, think that they have tremendous value because in performing experiments this way not only will your data give you a “best hit” but the spread of those data can tell you something about the underlying assumptions of your screen.
As part of my graduate work I designed and developed a screen for peptides that would activate only under a certain set of conditions. I could have done this by, for example, creating a peptide, testing it, and then make modifications to it and test again. This is a more, “rational design” approach and it’s a completely valid method. I’d invite you to think about the drawbacks of this method. How do I know how many times would I need to alter the peptide until it was perfect? Do I even know if my starting peptide is the right place to start? Instead of struggling with these questions I decided to utilize a “combinatorial method”. I synthesized a library of peptides that had varying controlled combinations of amino acids. The total number was over 10,000 different peptides yet the synthesis took only about two weeks! I was able to separate and test each peptide individually and I get a set of data similar to the one shown below (Figure 1). By setting the conditions X and Y I could find the peptides that had the max X response and the max Y response. These were my “positive hits”.
Figure 1 (Example of a “good” screen)
The spread of the data tells me how stringent my screening conditions are. If the conditions are too easy then no peptides are “selected for” in other words the data would look a lot like Figure 2. Can we really say whether or not any one of those dots is significantly different from any of the others? Of course, if the conditions are too harsh such that all the dots are zero then I won’t see any “positive hits” either. One other piece of information that can get overlooked is what the stringency tells us about the system our molecules live in. If X and Y are model biological conditions, perhaps these data suggest that our model is not close enough to reality. Alternatively, perhaps it means that the molecule we were already trying to force to evolve is already close to ideal. I think because of this it is shortsighted to say that combinatorial assays cannot address basic scientific questions.
Figure 2 (example of confusing data)
I hope to cover a whole host of different combinatorial assays in this blog. Not only is it a useful technique for peptides but also for studying the “yoga” of other biomolecules such as polymers and DNA. Additionally, these screens can end up being used in combination with advanced genetics techniques. If you have a question about combinatorial chemistry or a cool example please feel free to post it here for discussion! Thanks for reading!