Computational assessment of synthetic procedures.

Synthetic chemistry is hard because some reasonable looking molecules cannot be made, because there are errors in the chemical literature, because it is easy to miss reaction possibilities and because even the shape of molecules is very difficult to determine. We propose an approach to the computational analysis of reactions that tries to circumvent these difficulties, by restricting the analysis to simple rules for reactivity that can generate a large number of competing pathways. This huge ensemble is filtered using computational methods to pick out the most likely pathways, and to suggest possible products.