Prediction of Organic Reaction Outcomes Using Machine Learning

By Connor W. Coley, Regina Barzilay, Tommi S. Jaakkola, William H. Green Jr., Klavs F. Jensen. 2017

654 citations recorded by the source corpus.

Computer assistance in synthesis design has existed for over 40 years, yet retrosynthesis planning software has struggled to achieve widespread adoption. One critical challenge in developing high-quality pathway suggestions is that proposed reaction steps often fail when attempted in the laboratory, despite initially seeming viable. The true measure of success for any synthesis program is whether the predicted outcome matches what is observed experimentally. We report a model framework for anticipating reaction outcomes that combines the traditional use of reaction templates with the flexibility in pattern recognition afforded by neural networks. Using 15 000 experimental reaction records from granted United States patents, a model is trained to select the major (recorded) product by ranking a self-generated list of candidates where one candidate is known to be the major product. Candidate reactions are represented using a unique edit-based representation that emphasizes the fundamental transformation from reactants to products, rather than the constituent molecules’ overall structures. In a 5-fold cross-validation, the trained model assigns the major product rank 1 in 71.8% of cases, rank ≤3 in 86.7% of cases, and rank ≤5 in 90.8% of cases.

Subfields: Computational Theory and Mathematics, Materials Chemistry, Computational Chemistry & Drug Discovery

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Theory and Mathematics

In Materials Chemistry

In Computational Chemistry & Drug Discovery

Citations in the corpus

Planning chemical syntheses with deep neural networks and symbolic AI (2017, methodology): “[41, 42] Here, we make the assumption that the reactants can only react in the reported way.”
The need for uncertainty quantification in machine-assisted medical decision making (2019, background): “Some are in more applied areas such as drug discovery, while some are in more fundamental science areas such as the study of chemical reactions 12 , and assistance in the exploration and discovery of the molecular characteristics of medical phenomena from the available data using deep learning and…”
A robotic platform for flow synthesis of organic compounds informed by AI planning (2019, methodology): “A renewed in- terestinCASPbroughtaboutbyrecentadvancesindatascienceandmachinelearninghasledtosev-eralrecentstudies( 27 , 28 – 30 ), including Segler et al .”
“Found in Translation”: predicting outcomes of complex organic chemistry reactions using neural sequence-to-sequence models (2017, methodology): “Other approaches extract reaction templates directly from data [3–6].”
The Role of Machine Learning in the Understanding and Design of Materials (2020, background): “Data-driven approaches have recently been explored and shown to be promising for ﬁ nding the synthesis steps in organic retrosynthesis, 12,114,138,139 suggesting organic directing agents for the synthesis of zeolites, 140 and identifying new phases of inorganic compounds.”
Accelerating the discovery of materials for clean energy in the era of smart automation (2018, methodology): “Several recent studies 237–239 have involved training neural networks as a classifier to predict reaction outcomes on a database of known reactions.”
Unsupervised Attention-Guided Atom-Mapping (2020, methodology): “This match is called atom-mapping and is crucial for numerous tasks like template-based reaction prediction (Coley et al., 2017; Segler & Waller, 2017) and retrosynthesis planning methods (Segler et al., 2018; Thakkar et al., 2020; Fortunato et al., 2020), reaction graph neural network algorithms…”
GEOM: Energy-annotated molecular conformations for property prediction and molecular generation (2020, background): “Machine learning outperforms traditional rulebased baselines in many molecule-related tasks, including property prediction and virtual screening [1–3], inverse design using generative models [4–11], reinforcement learning [12–15], differentiable simulators [10, 16, 17], and synthesis planning and re”
State-of-the-art augmented NLP transformer models for direct and single-step retrosynthesis (2020, methodology): “The first retrosynthesis program LHASA [2] utilizes a template-based [3, 4] approach.”
Retrosynthetic Reaction Prediction Using Neural Sequence-to-Sequence Models (2017, methodology): “37 However, these types of deep learning approaches are fundamentally dependent on the rule-based expert system component and thus inherit some of its major limitations.”
Artificial intelligence in drug discovery: recent advances and future perspectives (2021, methodology): “With the aim of addressing some of these limitations, several steps have been taken to expand known reaction databases with negative reaction outcomes [177], and thereby, create new customized data compilations for automated synthesis planning [178].”
Augmenting large language models with chemistry tools (2023, background): “…a field, has been impacted through expert-designed artificial intelligence (AI) systems that tackle specific problems, such as reaction prediction 16–20 , retrosynthesis planning 21–27 , molecular property prediction 28–32 , de-novo molecular generation 33,34 , materials design 35,36 and, more…”