There's a good overview of DNA-encoded libraries from Raphael Franzini and Cassie Randolph from Utah that's worth a look for anyone wanting to quickly bring themselves up to speed on this promising new technology. As the article indicates, the whole field of DNA-encoded libraries, characterized by several related but distinct methods, has come of age in the sense of being transformed from a technology platform development paradigm to a more or less reliable springboard for screening large numbers of novel chemical entities against challenging protein targets.
The field is not free of roadblocks, however. As the authors reveal, much of encoded library technology runs into the same problems as traditional combinatorial libraries, only some of the problems are worse. Foremost among these are the size and molecular weight of the resulting molecules. The basic framework of a DNA-encoded library consists of 2, 3, 4 or more building blocks (BBs) that are strung together, either linearly or in a branched manner or in the form of a macrocycle. The BBs can be mono, di, trifunctional etc. and this feature directly impacts their availability, number and price as well as the versatility of the resulting library. It's not hard to see that you are going to get a rapid, almost exponential explosion in molecular weight with the number of building blocks, especially if you add in linkers connecting them together. Many of the millions or even billions of compounds emerging from these campaigns therefore are large, floppy. The problem is especially glaring for macrocycles; while there are very distinct advantages to them, the one big disadvantage is that you cannot easily take out a building block from a macrocycle without impacting its structural integrity, even if that building block is found to be extraneous to binding the target. The good news on the other hand is that the ClogP stays within reasonable limits in most cases.
The article explores other properties and challenges of encoded library technology. While the very high throughput and low cost enabled by cheap DNA sequencing make these libraries attractive, they are also plagued by problems of side reactions and impurities more often than you would like. These problems affect both the potency of the resulting molecules as well as subtleties in SAR, both of which are issues which you don't really want to deal with and can spend days pursuing. In addition, one big drawback of this technology is that because the chemistry used in them has to be compatible with DNA, the reaction space is quite limited relative to the medicinal chemist's toolkit: there's not much beyond amide coupling and click chemistry that you can use. Making more reaction space compatible with these libraries is definitely a big area for development.
Nonetheless, the article explores several novel chemotypes for promising targets emerging from screening such libraries which have been optimized into nanomolar hits and leads. In some cases the novelty of the chemotypes compares favorably with HTS campaigns. My one big question about the chemical matter from these libraries, and one which the review does not address, is the quality of the leads in terms of pharmacokinetic properties, especially cell permeability and clearance. Generally speaking it's going to be very hard to get large and floppy molecules past the cell membranes (cyclosporine notwithstanding, which seems to be very much of an exception in terms of its unique properties). In addition for something like a polyamide library, I would be worried about stability and clearance. Something tells me that it's going to be relatively easy to get potent binders from these libraries, but it's going to be much harder to turn them into bonafide drugs with good pharmacological properties than it is for 'ordinary' druglike molecules.
Notwithstanding the limitations, it's definitely true that all kinds of DNA-encoded libraries have now reached a stage where both small and big companies wanting to find hits for novel targets on the cheap can seriously think of at least wanting to collaborate with companies like GSK which have been longstanding players in the field. I would think the same promise and criticism would apply to outfits like XChem and Peptidream which are also using novel technology for generating combinatorial libraries. While the occasional success from these libraries would be welcome, you hope that whatever failings the libraries have would almost certainly be revealed in the ruthless attrition that drug discovery routinely faces. It's both the promise and curse of complex biology.
Sixty-four years later: How Watson and Crick did it
21 hours ago in The Curious Wavefunction