Chematica - A network model for chemistry

Scientists at Northwestern University have developed a beautiful network containing seven million chemical substances with the respective reactions that link them together. The software not only included the basic static data, but they are actually using machine learning applying over 86.000 chemical rules derived from existing knowlegde, which allow to explore relations between chemicals and help to identify the best paths to obtain a given chemical substance from others.

Having such a network of our knowledge of Chemistry can indeed be a great help to improve industrial processes and scientific research, since it tremendously simplifies the task of finding the best ways to synthesize chemicals or just find different ways to do the same thing. Using such a network model, it is possible to develop algorithms find the best path to synthesize a given chemical optimizing cost, ecological impact or available resources.

But this is not just an encyclopedia that allow us the explotation of existing knowledge and explore many connections which may be otherwise unknown without a tool like this, but also this kind of system will allow us to analyze the boundaries of our knowledge and project new possible compounds that are yet unknown.

Some interesting references...
The Guardian - Interview
The Grzybowski Group