If you'd like to check out what current chemical database files look like, "Enamine REAL" is a fairly widely-known one.[1] My understanding is that this file is

If you'd like to check out what current chemical database files look like, "Enamine REAL" is a fairly widely-known one.[1] My understanding is that this file is a mix of their ACTUAL in-stock inventory, as well as the product of running a small number of high-reliability reactions on each compound in that inventory. So it serves as a "vendor catalog" file, where everything in here can be ordered from Enamine and synthesized+delivered to your door in a few weeks. Another approach I've heard of for iterating through every molecule in a large region of chemical space is to START with a large molecule dataset, then for each molecule, predict the result of performing simple reactions on it. For each reaction product, do your full analysis, and only store the result if the analysis indicates it is noteworthy. This, in effect, lets you scan over a larger region of chemical space than you can fit in memory. [1] https://enamine.net/compound-collections/real-compounds/real-database https://enamine.net/compound-collections/real-compounds/real...