Retrosynthetic analysis

Retrosynthetic analysis is a technique for solving problems in the planning of organic syntheses. This is achieved by transforming a target molecule into simpler precursor structures regardless of any potential reactivity/interaction with reagents. Each precursor material is examined using the same method. This procedure is repeated until simple or commercially available structures are reached. These simpler/commercially available compounds can be used to form a synthesis of the target molecule. Retrosynthetic analysis was used as early as 1917 in Robinson's Tropinone total synthesis.^[1] Important conceptual work on retrosynthetic analysis was published by George Vladutz in 1963.^[2][3] E.J. Corey formalized and popularized the concept from 1967 onwards in his article General methods for the construction of complex molecules and his book The Logic of Chemical Synthesis.^[4][5][6][7]

The power of retrosynthetic analysis becomes evident in the design of a synthesis. The goal of retrosynthetic analysis is a structural simplification. Often, a synthesis will have more than one possible synthetic route. Retrosynthesis is well suited for discovering different synthetic routes and comparing them in a logical and straightforward fashion.^[8] A database may be consulted at each stage of the analysis, to determine whether a component already exists in the literature. In that case, no further exploration of that compound would be required. If that compound exists, it can be a jumping point for further steps developed to reach a synthesis.

There are both academic and commercial groups developing retrosynthesis tools. With the growing application of machine learning and artificial intelligence in chemistry, many research groups, such as the Coley Group from MIT, and companies, such as Chemical.AI, Reaxys, etc., have started to integrate deep learning into the conventional rule-based approaches.