Winnowing Down the Field of Molecules
How a medicinal chemist lights up the heavy-hitters and the clunkers
There are many, many challenges in any drug discovery programme. Just one of these is finding good molecular starting points that can be developed into a successful drug. High throughput screens (HTS) are a common method to search for these starting points. In an HTS, vast numbers of molecules are tested for interaction with a biological target of relevance to the disease of interest. At Charles River, our Early Discovery Biology department runs many high throughput screens each year for our clients. It’s my job to provide medicinal chemistry analysis on HTS projects with the goal of helping assess which molecules have the best potential to become drug candidates.
The reality of an HTS project is that the vast (on the order of hundreds of thousands of molecules) primary screen happens early on – after that there is still much work left to choose high quality molecules from the results. My colleagues in biology do a fantastic job at validating and miniaturising the assay, at running the screen quickly and efficiently, and at confirming the results in further rounds of testing. It's during these latter stages that I can have the most useful impact on the project – by helping tease the most promising signals from the background noise.
The primary screen often results in thousands of molecules that have shown a significant interaction with the biological target. To handle the large dataset, I have several powerful computational tools at my disposal. These include clustering molecules by similarities in their chemical structure, filtering molecules with less desirable functional groups and statistical methods to identify promiscuous molecules (those that interact with a great many targets and can have undesired biological effects). Together with my medicinal chemistry knowledge and intuition, I use these tools to highlight attractive and unattractive molecules and series of molecules from the dataset. An attractive molecule is one with significant potential to become an approved drug, but at these early stages of drug discovery the measure of that potential can be rather abstract and subjective. Some key features of an attractive molecule are potency (strength of interaction), selectivity against undesired targets, good physicochemical properties (that can lead to good solubility, permeability and stability) and synthetic feasibility (so that derivatisation shouldn’t add to the challenges of drug discovery). I also use visualisation software to examine overall trends and specific subsets within the dataset, and to create figures to illustrate the composition of the data and support my analysis. A simulated example of such a figure is shown below.
These analyses are immensely valuable to drug developers because it provides them with an impression of their dataset based on molecule structure and the potential of the molecules to be good drug discovery starting points. I can provide recommendations that augment those of the biology project leader on which compounds should progress to the next stage of the process on the basis of factors other than activity. Moreover, my analysis can, in tandem with all the further testing entailed, provide confidence in the validity of the HTS results.
The HTS projects I have been involved with have been very rewarding experiences. I enjoy working with and learning from my colleagues in biology. I really appreciate the opportunity to gain new perspectives by working on completely different targets for clients from around the world. As a medicinal chemist, the projects I work on routinely start from just a handful of molecules and our job is to create many new ones; I have relished the exciting new challenge of starting from hundreds of thousands of molecules and narrowing them down to just a handful of new ones.