RetroTransformDB: A Dataset of Generic Transforms for Retrosynthetic Analysis

Abstract

Presently, software tools for retrosynthetic analysis are widely used by organic, medicinal, and computational chemists. Rule-based systems extensively use collections of retro-reactions (transforms). While there are many public datasets with reactions in synthetic direction (usually non-generic reactions), there are no publicly-available databases with generic reactions in computer-readable format which can be used for the purposes of retrosynthetic analysis. Here we present RetroTransformDB-a dataset of transforms, compiled and coded in SMIRKS line notation by us. The collection is comprised of more than 100 records, with each one including the reaction name, SMIRKS linear notation, the functional group to be obtained, and the transform type classification. All SMIRKS transforms were tested syntactically, semantically, and from a chemical point of view in different software platforms. The overall dataset design and the retrosynthetic fitness were analyzed and curated by organic chemistry experts. The RetroTransformDB dataset may be used by open-source and commercial software packages, as well as chemoinformatics tools.

Autonomous Discovery in the Chemical Sciences Part I: Progress

“…[143] Generating yet-unseen chemical reactions for asynthesis plan-a necessity for the synthesis of novel molecules-is ah arder search problem than when searching within af ixed reaction network. [144] As the number of states in an aive retrosynthetic expansion will scale as b d for branching factor b and depth d,g uiding the search is an essential aspect of computer-aided synthesis planning (CASP) programs.T he breadth of the search depends on the coverage of the rule sets:A bstracted enzymatic reactions tend to number in the hundreds, [145] expert transformation rules often number in dozens or hundreds [146,147] but can extend into the tens of thousands in contemporary programs, [148] and algorithmically extracted templates generally number in the thousands to hundreds of thousands. [149][150][151][152] To the extent that reaction rules and synthetic strategies can be codified, synthesis planning is highly conducive to computational assistance [58,[153][154][155][156][157] (Figure 8).…”

Autonome Entdeckung in den chemischen Wissenschaften, Teil I: Fortschritt

“…[144] Da bei naiver retrosynthetischer Herangehensweise die Anzahl der Zustände mit b d skaliert, wobei b der Verzweigungsfaktor und d die Tiefe ist, stellt bei den Programmen zur rechnergestützten Syntheseplanung (Computer-Assisted-Synthesis Planning,C ASP) die Steuerung der Suche einen zentralen Aspekt dar. Der Umfang der Suche hängt von den anzuwendenden Regeln ab:A bstrahierte enzymatische Reaktionen gehen gerne in die Hunderte, [145] expertenbestimmte Umwandlungsregeln gibt es oft Dutzende oder Hunderte [146,147] (wobei es bei modernen Programmen bis Zehntausende sein kçnnen [148] ), und algorithmisch extrahierte Te mplate liegen im Allgemeinen im Bereich von Tausenden bis Hunderttausenden. [149][150][151][152] Soweit sich die Reaktionsregeln und Synthesestrategien kodifizieren lassen, kçnnen Rechnungen die Syntheseplanung stark unterstützen (Abbildung 8).…”

Autonomous discovery in the chemical sciences part I: Progress

“…Because the number of states in a naive retrosynthetic expansion will scale as b d for branching factor b and depth d, guiding the search is an essential aspect of computer-aided synthesis planning (CASP) programs. The breadth of the search depends on the coverage of the rule sets: abstracted enzymatic reactions tend to number in the hundreds [146], expert transformation rules often number in dozens or hundreds [147,148] but can extend into the tens of thousands in contemporary programs [149], and algorithmically-extracted templates generally number in the thousands to hundreds of thousands [150][151][152][153]. To the extent that reaction rules and synthetic strategies can be codified, synthesis planning is highly conducive to computational assistance [58,[154][155][156][157][158] (Figure 8).…”