In a famous footnote, the size of drug-like chemical space has been estimated to be 10^63 molecules. This number has remained controversial and merits re-investigation, but it is clear that the number of molecules with which we could, potentially, modulate biology far exceeds the number of molecules with which we now begin drug discovery campaigns. For instance, a large library screen for activity in an academic lab will rarely encompass more than 300,000 molecules, and even in Pharma this number rarely rises above 1.5 million.

A way to expand the range and number of compounds available for early discovery is to computationally screen libraries of compounds for fit to a biological structure, typically an enzyme or a receptor. This technology has been incubating for several decades, but has recently had key successes. In screens of 3 to 6 million molecules, "molecular docking" has discovered new chemical matter for multiple receptors, and found that the new compounds confer new pharmacology, often by triggering new signaling mechanisms.

Building on this work, we have sought to expand the libraries we are computationally docking towards 1 billion and, perhaps, 10 billion molecules. In exploratory studies we have modeled the diversity of the libraries and the chances finding new, active matter, with encouraging results. In actual testing of these large libraries against seven different targets, we have observed relatively high success rates, with potent, novel molecules, often with new signaling, emerging.

Open questions to be considered include:

* What is the pragmatic size of accessible, drug-like chemical space, and how does it scale?

* When does increasing library size become limited by our ability to prioritize and test candidates emerging from the computation?

* What improvements in theory and modeling will improve our success rates? What strategies can we use to maximize the impact of relatively cheap computation, and minimize the expense of relatively expensive experiment?

* Can we develop models for how large the biological opportunity is for novel compounds? How many different outcomes can one expect when targeting a particular receptor that can adopt a certain number of signaling-relevant states, with a certain number of likely off-targets, a certain number of plausible co-targets, and a certain number of pharmacological variables?

A short sketch of the current pipeline for compound advancement in drug discovery and chemical biology will be given to help set the context.