Pith. sign in

REVIEW 2 major objections 5 minor 1 cited by

Challenging reaction prediction models to generalize to novel chemistry

T0 review · 2 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

Pith's one-line read Random splits make reaction-prediction models look about 10 accuracy points better than they are, and the paper's split panel shows which reaction classes the models can and cannot extrapolate to.

desk verdict A careful empirical demonstration that random splits flatter reaction predictors; the NameRxn label-noise caveat is real but does not sink the central conclusion. read the letter →

arxiv 2501.06669 v1 pith:GHPD3SEF submitted 2025-01-11 cs.LG physics.chem-ph

classification cs.LGphysics.chem-ph
keywords reactionpredictionout-of-distributiongeneralizationdatasplitdesigndocument-leveltime-basedreaction-typeextrapolationNameRxntransformersequencemodel
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper argues that the standard way of benchmarking machine-learning reaction predictors—randomly splitting a reaction database into training and test sets—inflates reported accuracy, because a single patent or research team often contributes many near-identical reactions that end up on both sides of the split. To get realistic numbers, the authors evaluate a SMILES-based transformer on a cleaned 2.8-million-reaction version of the Pistachio database using splits that respect the data's actual structure: by document, by author, by publication year, and by entire reaction class. A model that scores 65% top-1 accuracy on a random split drops to 58% when whole documents are held out and 55% when whole authors are held out, and accuracy decays as the time gap between training data and test data widens. When entire reaction classes are withheld, the model extrapolates well to some classes (74–83% top-1 for specific Suzuki subtypes) and poorly to others (7–10% for Grignard ester and Heck), and the paper traces those failures to step composition and to stereochemistry and regiochemistry rather than to missing chemical knowledge. The paper's point is that reaction prediction is not as close to solved as leaderboard numbers suggest, and that this family of out-of-distribution evaluations should become the standard way to measure progress toward reaction discovery.

What carries the argument

The carrying object is a sequence of split constructions over a cleaned, deduplicated version of the Pistachio patent reaction database, each evaluated with the same encoder-decoder transformer (BART architecture with SMILES tokenization, closely related to the Molecular Transformer). Three split mechanisms do the principal work: document- and author-based splits, which stop a patent's series of closely related scope or structure-activity reactions from straddling the train/test boundary; a ladder of time cutoffs from 1996 to 2020 paired with held-out test sets for each year, making extrapolation distance a measurable axis; and NameRxn reaction-class splits, which hold out complete classes (Grignard ester, Heck, Chloro Suzuki, Triflyloxy Suzuki, and all Suzuki) and contrast them with a baseline given 1,000 in-class examples, separating a class's intrinsic difficulty from its extrapolation difficulty. A diagnostic supplement computes fingerprint distances from each test reaction to its five nearest training neighbors in reactant space and in reaction-transform space, which distinguishes splits that shift the substrate distribution from splits that shift the transform distribution and correlates with the observed difficulty.

What would settle it

Re-run the All Suzuki holdout with an independent, mechanism-based classifier that identifies Suzuki couplings by bond changes and catalyst rather than by the vendor label; if the model still scores near 51% top-1 after every mechanistically equivalent reaction is provably absent from training, the extrapolation claim stands, whereas a sharp accuracy jump after purging mislabeled or unclassified examples would show the result was partly leakage.

Watch

Extended reading notes

Core claim

The central claim, on the paper's own terms, is that accuracy figures reported on standard in-distribution benchmarks are over-optimistic, and that a panel of out-of-distribution splits—by document, by author, by time, and by reaction class—gives a more faithful measure of how reaction predictors behave in deployment and whether they can extrapolate to novel chemistry. A BART-based encoder-decoder transformer, close in design to the Molecular Transformer, reaches 65% top-1 accuracy on a random reaction-level split of Pistachio patent data but only 58% on a document-level split and 55% on an author-level split. On time-based splits, accuracy declines as the extrapolation distance grows, yet a model trained only on reactions reported through 1996 still makes correct top-1 predictions some 25 years out; on a static held-out set of Buchwald–Hartwig reactions its top-1 accuracy climbs from 5% to above 60% as later training cutoffs add more in-class examples. On reaction-type splits, held-out Chloro Suzuki and Triflyloxy Suzuki classes are extrapolated with less than a 10-point drop (74% and 83% top-1), largely because other Suzuki reactions remain in training, whereas removing the whole Suzuki family cuts All Suzuki accuracy to 51%. Grignard ester and Heck holdouts are far harder (10% and 7% top-1), and the paper shows these failures are partly compositional and representational: iterating the Grignard double addition as two single steps raises accuracy from 4.5% to 49.7%, and ignoring stereochemistry and then regiochemistry raises the Heck extrapolation score from 7% to 42%.

Load-bearing premise

Every reaction-type result assumes NameRxn tagging is complete, so that removing the named classes plus all uncategorized reactions strips every training example of the held-out transform; if reactions that are mechanistically the same survive in training under another label, the reported extrapolation accuracies such as 51% top-1 on all Suzuki are inflated.

Editorial extensions

If this is right

  • Random-split leaderboard numbers overstate deployment accuracy by roughly 7–10 top-1 points, so future reaction-prediction papers should report document- and author-level splits alongside them.
  • Time-based evaluation bounds a model's useful lifetime: accuracy decays with the gap between training cutoff and test year, but the non-zero accuracy of a 1996-trained model decades later shows future reactions are partly predictable from past chemistry.
  • Extrapolation to an unseen reaction class works when a sibling family stays in training (Chloro and Triflyloxy Suzuki) and largely fails when the whole family is removed (All Suzuki, 51% top-1), which means claims about reaction discovery must be tested with full-family holdouts.
  • Much of the observed failure is recoverable: iterating the Grignard double-addition prediction as two single steps raises accuracy from 4.5% to 49.7%, and evaluation that ignores stereochemistry and then regiochemistry raises Heck extrapolation from 7% to 42%.
  • The three split families answer different questions—retrospective deployment, prospective use, and reaction discovery—so they should be reported together rather than treated as interchangeable benchmarks.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The split construction is author- and document-dense by design, so the 10-point gap probably understates the inflation of random splits on datasets with broader author diversity, such as academic literature or high-throughput experimentation collections.
  • The Grignard double-addition result suggests a general design principle the paper leaves implicit: reaction predictors may gain more from a step-decomposition objective, in which the model predicts one elementary transform and feeds the product forward, than from more training data.
  • The nearest-neighbor fingerprint distances offer a cheap pre-registration rule for future benchmarks: large reaction-transform distances should predict behavior like the Grignard ester, Heck, and All Suzuki cases, while large reactant distances should predict behavior like the document and author splits.
  • A natural next experiment not performed here is to run the same split panel on graph-based or pretrained-language-model reaction predictors; the ordering of which classes extrapolate well could serve as a fingerprint of a model's inductive bias.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

2 major / 5 minor

Summary. The paper evaluates a BART-based encoder-decoder transformer for reaction product prediction on a cleaned, deduplicated subset of the Pistachio patent dataset, comparing several out-of-distribution (OOD) split strategies against the standard random split. It reports that random splits overestimate top-1 accuracy (65%) relative to document-based (58%) and author-based (55%) splits; that time-based splits show accuracy declining with extrapolation distance from the training cutoff; and that reaction-type splits based on NameRxn classes give mixed evidence about zero-shot extrapolation, with Suzuki subtypes being easier than Grignard ester or Heck reactions. The authors further analyze why certain classes are harder, such as the need for multi-step reasoning in Grignard double additions and stereochemical/regiochemical sensitivity in Heck reactions, and they provide a fingerprint-distance analysis of the distribution shifts induced by each split type. The paper explicitly positions itself as a benchmark and evaluation study, releasing code for data cleaning and model training.

Significance. If the findings hold, the paper makes a useful and timely contribution by quantifying how much standard random splits overstate reaction-prediction performance and by proposing a suite of complementary OOD splits (document, author, time, reaction-type) that better reflect deployment. The careful dataset cleaning, the controlled training-size comparisons, and the fingerprint-based analysis of split-induced distribution shifts are strengths, as is the public release of code. The time-split protocol with multiple cutoffs and the Buchwald–Hartwig case study are particularly nice. However, the reaction-type extrapolation results—which underpin the paper's statements about reaction discovery—depend on the accuracy and completeness of NameRxn labels, and the paper does not validate those labels. The study also uses a single model and single training runs, so the quantitative accuracy gaps should be interpreted with appropriate caution.

major comments (2)
  1. [S1.4, Fig. 4, Table S2.4] The reaction-type splits, especially the 'All Suzuki' zero-shot evaluation, are not robust to NameRxn label noise. The paper removes only uncategorized reactions (NameRxn '0.0') before forming splits, explicitly acknowledging this is to 'avoid inadvertently training on reactions that might be similar to those that we are trying to exclude.' However, NameRxn covers only 74% of the cleaned dataset (stated in S1.4), so a Suzuki reaction mislabeled as, say, a generic cross-coupling or another named class would remain in the 1M-reaction training set. This could inflate the reported extrapolation accuracy (e.g., 51% top-1 for All Suzuki) and weaken the paper's conclusion that models possess some ability to extrapolate to unseen reaction types. The same issue affects the Buchwald–Hartwig test set in S1.3, which is constructed from NameRxn codes 1.3.1–1.3.4 and 1.9.43; BH reactions mislabeled under other codes could appear in the training sets of the time-split models. To support the central claims of Section 2.3 and Section 2.2.1, the authors should validate NameRxn assignments on a representative sample, or use an independent reaction classifier (e.g., template matching or mechanistic rules) to detect and remove reactions of the held-out type from the training set, then re-evaluate. A sensitivity analysis that relaxes the definition of the held-out class would also help establish how much the results depend on label completeness.
  2. [Section 2.1, Fig. 2, Table S2.2; Section 2.2, Fig. 3] The headline accuracy numbers are reported from single training runs with no uncertainty intervals. The gap between random, document, and author splits (65%, 58%, 55%) is large enough that it is probably real, but the time-split results in Figure 3A show substantial inter-year variability, and the paper itself attributes part of this to 'variability from the random selection of test sets.' Without multiple seeds, standard errors, or confidence intervals, it is difficult to assess whether smaller differences—for example between the document- and author-based splits, or between adjacent time cutoffs—are meaningful. Since the paper's central claim is that random splits overestimate performance, and since it makes quantitative claims such as 'the drop of ~10% accuracy,' the authors should provide at least bootstrap confidence intervals over the test set, or ideally repeat training with a few seeds for the main splits. This would not change the qualitative direction of the results, but it would make the reported magnitudes more interpretable.
minor comments (5)
  1. [Section 2.1, first paragraph] The phrase 'a model trained and evaluated on an on reactions split' is grammatically awkward and should be rephrased, e.g., 'a model trained and evaluated on a reaction-level (random) split.'
  2. [Figure 3A caption] The description of the dashed line is confusing: the caption says 'the dashed line indicates model performance when the model is extrapolating,' but in the figure the dashed line appears to be a boundary separating interpolative and extrapolative regions. Please clarify what the dashed line represents and how the reader is to interpret the region to its right.
  3. [Figure 5B and 5C] The small SMILES structures and the labels 'True' and 'False' are not self-explanatory; the reader has to infer that 'True' is the correct product and 'False' is an incorrect model prediction. A legend or a sentence in the caption would help.
  4. [S1.6] The reaction fingerprint is defined as the difference between product and reactant fingerprints, but the text then refers to 'cosine distance' between these difference vectors; it would be clearer to state explicitly that the cosine distance is computed on these difference vectors.
  5. [S1.4] The paper notes that the NameRxn system is not the only possible classification scheme and that different schemes may lead to different results; this is a good caveat, but it would be even more informative to include one such alternative (e.g., RXNO or template-based classes) for at least one held-out class, to test the sensitivity of the extrapolation conclusions to the choice of taxonomy.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper's claims are empirical measurements from explicitly defined dataset splits, with no fitted parameter renamed as a prediction and no load-bearing self-citation chain.

full rationale

The paper's central results are direct empirical comparisons: accuracies under reaction-, document-, and author-based splits (Section 2.1), time-based extrapolation curves (Section 2.2), and held-out reaction-class accuracies (Section 2.3). Each split is defined by external metadata or the NameRxn classification system, not by model predictions or by quantities fitted to the evaluation outcome. Hyperparameters are tuned on validation sets and held fixed for the OOD evaluations, which is standard practice and does not make the reported accuracies circular. The Buchwald–Hartwig and NameRxn test sets are constructed by selecting reactions with specified NameRxn codes, and the reported accuracies are simply measured on those sets; no step derives a prediction from its own input by construction. Self-citations (e.g., references to prior work by the same authors on generative models, retrosynthesis, or reaction discovery) appear only as background, tooling, or baseline context, and none is invoked as a uniqueness theorem or as the justification for a forbidden alternative. The reviewer-identified concern about NameRxn label noise potentially inflating the 'All Suzuki' zero-shot accuracy is a data-quality and external-validity limitation, not a circularity: the split definition does not assume the labels are correct, and the model does not contribute to those labels. Overall, the derivation chain is self-contained and the findings stand as empirical evidence independent of the paper's own assumptions.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The paper does not introduce a derivation; its results are empirical and depend on dataset choices and domain assumptions rather than free mathematical constants. The central empirical claims are robust to reasonable variations, but the quantitative magnitudes depend on cleaning thresholds, split budgets, the proprietary Pistachio source, and the chosen model.

free parameters (4)
  • Dataset cleaning thresholds = Heavy atoms >=5; product heavy atoms >=2; token length <=800; etc.
    S1.1 hand-set filters define which reactions enter the dataset and therefore affect the difficulty of every split.
  • Split size budgets = 1M train, 100k test, 30k/10k/2k validation, 3k test per year, 1k baseline OOD sample
    S1.2-S1.4 choose these sizes by hand; they set the statistical power of each evaluation and influence confidence in the reported accuracies.
  • Nearest-neighbor count k = 5
    S1.6 uses average cosine distance to the 5 nearest training neighbors in Fig. 6; the choice of k is arbitrary.
  • Model hyperparameters = Selected by HPO over 100 trials per task (e.g., learning rate 1e-5 to 1e-2, encoder/decoder layers 2-12)
    S1.5 tunes hyperparameters on validation splits; exact point values are not reported, but they affect the absolute accuracy of the evaluated model.
assumptions (5)
  • domain assumption Pistachio 2022Q4 US grants reactions and their metadata (author, document, year, NameRxn class) are a faithful sample of real reaction chemistry.
    Invoked throughout S1.1-S1.4; if the database or metadata is biased, all OOD estimates inherit that bias. The authors note in the Conclusion that only patent data are considered.
  • domain assumption The cleaning and deduplication steps in S1.1 remove all identical reactions across training and test splits.
    S1.1 says deduplication 'ensures that at test time we do not evaluate on an example that completely matches one in the training set.' If incomplete, reported accuracies would be inflated.
  • domain assumption NameRxn class labels are accurate enough that removing a class and the uncategorized reactions removes essentially all training examples of that transform.
    S1.4; this is the load-bearing premise for the reaction-type splits; classification noise would leak held-out reactions into training and inflate extrapolation accuracy.
  • domain assumption Morgan fingerprint cosine distance is a valid proxy for chemical and reaction similarity.
    S1.6; used to characterize how the OOD splits differ. The authors call fingerprints a quick-to-compute practical guide, not a complete picture.
  • domain assumption A single BART-style Transformer is representative of current reaction predictors for drawing qualitative conclusions.
    S1.5; the paper explicitly says the model likely is not state of the art and was chosen for tractability. Other model families could show different drop magnitudes.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Challenging reaction prediction models to generalize to novel chemistry." pith.science (2026). https://pith.science/paper/GHPD3SEF

@misc{pith2026250106669,
  author       = {Pith},
  title        = {Pith review of: Challenging reaction prediction models to generalize to novel chemistry},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/GHPD3SEF}},
  note         = {Machine review of arXiv:2501.06669}
}
read the original abstract

Deep learning models for anticipating the products of organic reactions have found many use cases, including validating retrosynthetic pathways and constraining synthesis-based molecular design tools. Despite compelling performance on popular benchmark tasks, strange and erroneous predictions sometimes ensue when using these models in practice. The core issue is that common benchmarks test models in an in-distribution setting, whereas many real-world uses for these models are in out-of-distribution settings and require a greater degree of extrapolation. To better understand how current reaction predictors work in out-of-distribution domains, we report a series of more challenging evaluations of a prototypical SMILES-based deep learning model. First, we illustrate how performance on randomly sampled datasets is overly optimistic compared to performance when generalizing to new patents or new authors. Second, we conduct time splits that evaluate how models perform when tested on reactions published in years after those in their training set, mimicking real-world deployment. Finally, we consider extrapolation across reaction classes to reflect what would be required for the discovery of novel reaction types. This panel of tasks can reveal the capabilities and limitations of today's reaction predictors, acting as a crucial first step in the development of tomorrow's next-generation models capable of reaction discovery.

Figures

Figures reproduced from arXiv: 2501.06669 by the authors.

Figure 1
Figure 1. (A) Reaction prediction, in the context of this manuscript, is the task of predicting the major product(s) of a reaction given the reactants. (Note that by “reaction” we mean specific reported reaction examples, rather than generic reaction “types” or “classes” that cover a large group of related specific examples—we will come back to the concept of reaction types in Section 2.3.) Chemical reaction datasets are ofte… view at source ↗
Figure 2
Figure 2. (A) Reaction datasets are formed by authors coming together and writing documents, which contain many (often similar) reactions. Evaluating a reaction predictor on train/test sets that account for this structure provide different accuracy scores. (Note that in this paper we clean and deduplicate reactions before creating the splits, such that a reaction is only associated with one document—see Section S1.1.) (B) Top… view at source ↗
Figure 3
Figure 3. (A) Top-1 accuracies of reaction predictors trained up to different time cutoffs (different colors) when evaluated on held-out test sets for each year (x-axis). For instance, the line in the lightest shade, marked “1996”, reports the top-1 accuracy for a reaction predictor trained on reactions that were reported up to 1996 (inclusive). The dashed line indicates model performance when the model is “extrapolating”—mea… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Top-1, 3, and 5 accuracies for reaction predictors evaluated on different reaction-type splits. Each column [PITH_FULL_IMAGE:figures/full_fig_p006_4.png]
Figure 5
Figure 5. Figure 5: We investigate reasons for the contrasting performance in the different NameRxn splits. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Distribution of the average cosine distance between each reaction in each test set to its nearest five [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Retro-Rank-In: A Ranking-Based Approach for Inorganic Materials Synthesis Planning

    physics.chem-ph 2025-02 conditional novelty 7.0 of 10

    A pairwise ranking model over a shared embedding space outperforms prior fixed-vocabulary classifiers for inorganic retrosynthesis, especially when ranking many candidate precursor sets.

Reference graph

Works this paper leans on

108 extracted references · 54 canonical work pages · cited by 1 Pith paper

  1. [1]

    Reagent prediction with a molecular transformer improves reaction data quality

    Mikhail Andronov, Varvara Voinarovska, Natalia Andronova, Michael Wand, Djork-Arné Clevert, and Jürgen Schmidhuber. Reagent prediction with a molecular transformer improves reaction data quality. Chemical Science, 14(12):3235–3246, 2023. doi:10.1039/D2SC06798F

  2. [2]

    ASKCOS (Automated System for Knowledge-based Continuous Organic Synthesis), 2019

    ASKCOS Team. ASKCOS (Automated System for Knowledge-based Continuous Organic Synthesis), 2019. URL https://askcos.mit.edu/

  3. [3]

    Learning to Split for Automatic Bias Detection

    Yujia Bao and Regina Barzilay. Learning to split for automatic bias detection. arXiv [cs.LG] , 2022. doi:10.48550/arXiv.2204.13749

  4. [4]

    Non-autoregressive electron redistribution modeling for reaction prediction

    Hangrui Bi, Hengyi Wang, Chence Shi, Connor W Coley, Jian Tang, and Hongyu Guo. Non-autoregressive electron redistribution modeling for reaction prediction. In Proceedings of the 38th International Conference on Machine Learning, volume 139, pages 904–913. PMLR, 2021

  5. [5]

    Discovery of novel chemical reactions by deep generative recurrent neural network

    William Bort, Igor I Baskin, Timur Gimadiev, Artem Mukanov, Ramil Nugmanov, Pavel Sidorov, Gilles Marcou, Dragos Horvath, Olga Klimchuk, Timur Madzhidov, and Alexandre Varnek. Discovery of novel chemical reactions by deep generative recurrent neural network. Scientific reports, 11(1):3178, 2021. doi:10.1038/s41598- 021-81889-y

  6. [6]

    A generative model for electron paths

    John Bradshaw, Matt J Kusner, Brooks Paige, Marwin H S Segler, and José Miguel Hernández-Lobato. A generative model for electron paths. In International Conference on Learning Representations 2019, 2019

  7. [7]

    A model to search for synthesizable molecules

    John Bradshaw, Brooks Paige, Matt J Kusner, Marwin H S Segler, and José Miguel Hernández-Lobato. A model to search for synthesizable molecules. In Advances in Neural Information Processing Systems 32, pages 7937–7949. Curran Associates, Inc., 2019

  8. [8]

    Barking up the right tree: an approach to search over molecule synthesis DAGs

    John Bradshaw, Brooks Paige, Matt J Kusner, Marwin H S Segler, and José Miguel Hernández-Lobato. Barking up the right tree: an approach to search over molecule synthesis DAGs. In Advances in Neural Information Processing Systems 33. Curran Associates Inc., 2020

Show all 108 references
  1. [9]

    Analysis of the reactions used for the preparation of drug candidate molecules

    John S Carey, David Laffan, Colin Thomson, and Mike T Williams. Analysis of the reactions used for the preparation of drug candidate molecules. Organic & biomolecular chemistry , 4(12):2337–2347, 2006. doi:10.1039/b602413k

  2. [10]

    On evaluating adversarial robustness

    Nicholas Carlini, Anish Athalye, Nicolas Papernot, Wieland Brendel, Jonas Rauber, Dimitris Tsipras, Ian Good- fellow, Aleksander Madry, and Alexey Kurakin. On evaluating adversarial robustness. arXiv [cs.LG], 2019. doi:10.48550/arXiv.1902.06705

  3. [11]

    Assessing the extrapolation capability of template-free retrosynthesis models

    Shuan Chen and Yousung Jung. Assessing the extrapolation capability of template-free retrosynthesis models. arXiv [physics.chem-ph], 2024. doi:10.48550/arXiv.2403.03960

  4. [12]

    SCUBIDOO: A large yet screenable and easily searchable database of computationally created chemical compounds optimized toward high likelihood of synthetic tractability

    F Chevillard and P Kolb. SCUBIDOO: A large yet screenable and easily searchable database of computationally created chemical compounds optimized toward high likelihood of synthetic tractability. Journal of chemical information and modeling, 55(9):1824–1835, 2015. doi:10.1021/a...

  5. [13]

    ChemBERTa: Large-scale self-supervised pretraining for molecular property prediction

    Seyone Chithrananda, Gabriel Grand, and Bharath Ramsundar. ChemBERTa: Large-scale self-supervised pretraining for molecular property prediction. In Machine Learning for Molecules Workshop at NeurIPS 2020,

  6. [14]

    A graph-convolutional neural network model for the prediction of chemical reactivity

    Connor W Coley, Wengong Jin, Luke Rogers, Timothy F Jamison, Tommi S Jaakkola, William H Green, Regina Barzilay, and Klavs F Jensen. A graph-convolutional neural network model for the prediction of chemical reactivity. Chemical science, 10(2):370–377, 2019. doi:10.1039/C8SC04228D

  7. [15]

    Graph transformation policy network for chemical reaction prediction

    Kien Do, Truyen Tran, and Svetha Venkatesh. Graph transformation policy network for chemical reaction prediction. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, page 750–760, 2019. doi:10.1145/3292500.3330958

  8. [16]

    Deep learning for chemical reaction prediction.Molecular Systems Design & Engineering, 3(3):442–452, 2018

    David Fooshee, Aaron Mood, Eugene Gutman, Mohammadamin Tavakoli, Gregor Urban, Frances Liu, Nancy Huynh, David Van Vranken, and Pierre Baldi. Deep learning for chemical reaction prediction.Molecular Systems Design & Engineering, 3(3):442–452, 2018. doi:10.1039/C7ME00107J

  9. [17]

    The 25th anniversary of the Buchwald–Hartwig amination: Development, applications, and outlook

    Paola A Forero-Cortés and Alexander M Haydl. The 25th anniversary of the Buchwald–Hartwig amination: Development, applications, and outlook. Organic process research & development , 23(8):1478–1483, 2019. doi:10.1021/acs.oprd.9b00161

  10. [18]

    Amortized tree generation for bottom-up synthesis planning and synthesizable molecular design

    Wenhao Gao, Rocío Mercado, and Connor W Coley. Amortized tree generation for bottom-up synthesis planning and synthesizable molecular design. In International Conference on Learning Representations 2022, 2022

  11. [19]

    Holistic chemical evaluation reveals pitfalls in reaction prediction models

    Victor Sabanza Gil, Andrés M Bran, Malte Franke, Remi Schlama, J Luterbacher, and Philippe Schwaller. Holistic chemical evaluation reveals pitfalls in reaction prediction models. arXiv [physics.chem-ph], 2023. doi:10.48550/arXiv.2312.09004

  12. [20]

    Deep Learning

    Ian Goodfellow, Yoshua Bengio, and Aaron Courville. Deep Learning. The MIT Press, 2016

  13. [21]

    Explaining and harnessing adversarial examples

    Ian J Goodfellow, Jonathon Shlens, and Christian Szegedy. Explaining and harnessing adversarial examples. arXiv [stat.ML], 2014. doi:10.48550/arXiv.1412.6572

  14. [22]

    Reaction prediction and synthesis design

    Jonathan M Goodman. Reaction prediction and synthesis design. In Applied Chemoinformatics, pages 86–105. Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim, Germany, 2018. doi:10.1002/9783527806539.ch4b

  15. [23]

    Learning to navigate the synthetically accessible chemical space using reinforcement learning

    Sai Krishna Gottipati, Boris Sattarov, Sufeng Niu, Yashaswi Pathak, Haoran Wei, Shengchao Liu, Karam M J Thomas, Simon Blackburn, Connor W Coley, Jian Tang, Sarath Chandar, and Yoshua Bengio. Learning to navigate the synthetically accessible chemical space using reinforcement ...

  16. [24]

    Dataset bias in the natural sciences: A case study in chemical reaction prediction and synthesis design

    Ryan-Rhys Griffiths, Philippe Schwaller, and Alpha Lee. Dataset bias in the natural sciences: A case study in chemical reaction prediction and synthesis design. ChemRxiv, 2018. doi:10.26434/chemrxiv.7366973.v1

  17. [25]

    In search of lost domain generalization

    Ishaan Gulrajani and David Lopez-Paz. In search of lost domain generalization. In International Conference on Learning Representations 2020, 2020

  18. [26]

    Palladium-catalyzed aromatic aminations with in situ generated aminostannanes

    Anil S Guram and Stephen L Buchwald. Palladium-catalyzed aromatic aminations with in situ generated aminostannanes. Journal of the American Chemical Society, 116(17):7901–7902, 1994. doi:10.1021/ja00096a059

  19. [27]

    A baseline for detecting misclassified and out-of-distribution examples in neural networks

    Dan Hendrycks and Kevin Gimpel. A baseline for detecting misclassified and out-of-distribution examples in neural networks. In International Conference on Learning Representations 2017, 2017

  20. [28]

    Reaction planning: prediction of new organic reactions

    Rainer Herges. Reaction planning: prediction of new organic reactions. Journal of Chemical Information and Computer Sciences, 30(4):377–383, 1990. doi:10.1021/ci00068a006

  21. [29]

    Molecular design in synthetically accessible chemical space via deep reinforcement learning

    Julien Horwood and Emmanuel Noutahi. Molecular design in synthetically accessible chemical space via deep reinforcement learning. ACS omega, 5(51):32984–32994, 2020. doi:10.1021/acsomega.0c04153. 12

  22. [30]

    LoRA: Low-rank adaptation of large language models

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. LoRA: Low-rank adaptation of large language models. In International Conference on Learning Represen- tations 2022, 2022

  23. [31]

    Chemformer: A pre-trained transformer for computational chemistry

    Ross Irwin, Spyridon Dimitriadis, Jiazhen He, and Esben Jannik Bjerrum. Chemformer: A pre-trained transformer for computational chemistry. Machine Learning: Science and Technology , 3(1):015022, 2022. doi:10.1088/2632-2153/ac3ffb

  24. [32]

    Transformer performance for chemical reactions: Analysis of different predictive and evaluation scenarios

    Fernando Jaume-Santero, Alban Bornet, Alain Valery, Nona Naderi, David Vicente Alvarez, Dimitrios Proios, Anthony Yazdani, Colin Bournez, Thomas Fessard, and Douglas Teodoro. Transformer performance for chemical reactions: Analysis of different predictive and evaluation scenar...

  25. [33]

    Predicting organic reaction outcomes with Weisfeiler-Lehman network

    Wengong Jin, Connor W Coley, Regina Barzilay, and Tommi Jaakkola. Predicting organic reaction outcomes with Weisfeiler-Lehman network. In Advances in Neural Information Processing Systems 30, pages 2607–2616. Curran Associates, Inc., 2017

  26. [34]

    Latent biases in machine learning models for predicting binding affinities using popular data sets

    Ganesh Chandan Kanakala, Rishal Aggarwal, Divya Nayar, and U Deva Priyakumar. Latent biases in machine learning models for predicting binding affinities using popular data sets. ACS omega, 8(2):2389–2397, 2023. doi:10.1021/acsomega.2c06781

  27. [35]

    ReactionPredictor: prediction of complex chemical reactions at the mechanistic level using machine learning

    Matthew A Kayala and Pierre Baldi. ReactionPredictor: prediction of complex chemical reactions at the mechanistic level using machine learning. Journal of chemical information and modeling, 52(10):2526–2540,

  28. [36]

    Pursuing a prospective perspective

    Steven Kearnes. Pursuing a prospective perspective. Trends in chemistry , 3(2):77–79, 2021. doi:10.1016/j.trechm.2020.10.012

  29. [37]

    WILDS: A benchmark of in-the-wild distribution shifts

    Pang Wei Koh, Shiori Sagawa, Henrik Marklund, Sang Michael Xie, Marvin Zhang, Akshay Balsubramani, Weihua Hu, Michihiro Yasunaga, Richard Lanas Phillips, Irena Gao, Tony Lee, Etienne David, Ian Stavness, Wei Guo, Berton A Earnshaw, Imran S Haque, Sara Beery, Jure Leskovec, Ans...

  30. [38]

    ChemBO: Bayesian optimization of small organic molecules with synthesizable recommendations

    Ksenia Korovina, Sailun Xu, Kirthevasan Kandasamy, Willie Neiswanger, Barnabas Poczos, Jeff Schneider, and Eric P Xing. ChemBO: Bayesian optimization of small organic molecules with synthesizable recommendations. In Proceedings of the Twenty Third International Conference on A...

  31. [39]

    Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias

    Dávid Péter Kovács, William McCorkindale, and Alpha A Lee. Quantitative interpretation explains machine learning models for chemical reaction prediction and uncovers bias. Nature communications, 12(1):1695, 2021. doi:10.1038/s41467-021-21895-w

  32. [40]

    NameRxn: More than just a reaction classifier

    Ingvar Lagerstedt, John Mayfield, and Roger Sayle. NameRxn: More than just a reaction classifier. ACS Fall 2021,

  33. [41]

    Simple and scalable predictive uncertainty estimation using deep ensembles

    Balaji Lakshminarayanan, Alexander Pritzel, and Charles Blundell. Simple and scalable predictive uncertainty estimation using deep ensembles. In Advances in Neural Information Processing Systems 30, pages 6402–6413. Curran Associates, Inc., 2017

  34. [42]

    SIMPD: An algorithm for generating simulated time splits for validating machine learning approaches

    Gregory A Landrum, Maximilian Beckers, Jessica Lanini, Nadine Schneider, Nikolaus Stiefl, and Sereina Riniker. SIMPD: An algorithm for generating simulated time splits for validating machine learning approaches. Journal of cheminformatics, 15(1):119, 2023. doi:10.1186/s13321-0...

  35. [43]

    Handwritten digit recognition with a back-propagation network

    Yann LeCun, Bernhard E Boser, John S Denker, Donnie Henderson, R E Howard, Wayne E Hubbard, and Lawrence D Jackel. Handwritten digit recognition with a back-propagation network. In Advances in Neural Information Processing Systems 2, pages 396–404. Morgan-Kaufmann, 1990. 13

  36. [44]

    BART: Denoising sequence-to-sequence pre-training for natural language genera- tion, translation, and comprehension

    Mike Lewis, Yinhan Liu, Naman Goyal, Marjan Ghazvininejad, Abdelrahman Mohamed, Omer Levy, Ves Stoy- anov, and Luke Zettlemoyer. BART: Denoising sequence-to-sequence pre-training for natural language genera- tion, translation, and comprehension. In Proceedings of the 58th Annu...

  37. [45]

    A system for massively parallel hyperparameter tuning

    Liam Li, Kevin Jamieson, Afshin Rostamizadeh, Ekaterina Gonina, Jonathan Ben-tzur, Moritz Hardt, Benjamin Recht, and Ameet Talwalkar. A system for massively parallel hyperparameter tuning. In Proceedings of Machine Learning and Systems, volume 2, pages 230–246, 2020

  38. [46]

    Decoupled weight decay regularization

    Ilya Loshchilov and Frank Hutter. Decoupled weight decay regularization. In International Conference on Learning Representations 2019, 2019

  39. [47]

    Extraction of chemical structures and reactions from the literature

    Daniel Mark Lowe. Extraction of chemical structures and reactions from the literature. PhD thesis, University of Cambridge, 2012. URL http://dx.doi.org/10.17863/CAM.16293

  40. [48]

    Ideation and evaluation of novel multicomponent reactions via mechanistic network analysis and automation

    Babak Mahjour, Juncheng Lu, Jenna Fromer, Nicholas Casetti, and Connor Coley. Ideation and evaluation of novel multicomponent reactions via mechanistic network analysis and automation. ChemRxiv, 2024. doi:10.26434/chemrxiv-2024-qfjh9-v3

  41. [49]

    Pistachio - search and faceting of large reaction databases

    John Mayfield, Daniel Lowe, and Roger Sayle. Pistachio - search and faceting of large reaction databases. ACS Fall 2017, 2017. URL https://nextmovesoftware.com/talks/Mayfield_Pistachio_NIHReactions_202105. pdf

  42. [50]

    Doubly stochastic graph-based non-autoregressive reaction prediction

    Ziqiao Meng, Peilin Zhao, Yang Yu, and Irwin King. Doubly stochastic graph-based non-autoregressive reaction prediction. In Proceedings of the Thirty-Second International Joint Conference on Artificial Intelligence Main Track., pages 4064–4072, 2023. doi:10.24963/ijcai.2023/452

  43. [51]

    Ray: A distributed framework for emerging AI applications

    Philipp Moritz, Robert Nishihara, Stephanie Wang, Alexey Tumanov, Richard Liaw, Eric Liang, Melih Elibol, Zongheng Yang, William Paul, Michael I Jordan, and Ion Stoica. Ray: A distributed framework for emerging AI applications. In 13th USENIX Symposium on Operating Systems Des...

  44. [52]

    Pistachio

    NextMove Software. Pistachio. https://www.nextmovesoftware.com/pistachio.html, 2021. Accessed: 2021-11-18

  45. [53]

    NameRxn (expert system for named reaction identification and classification), 2022

    NextMove Software. NameRxn (expert system for named reaction identification and classification), 2022. URL https://www.nextmovesoftware.com/namerxn.html

  46. [54]

    Palladium-catalyzed formation of carbon-nitrogen bonds

    Frederic Paul, Joe Patt, and John F Hartwig. Palladium-catalyzed formation of carbon-nitrogen bonds. reaction intermediates and catalyst improvements in the hetero cross-coupling of aryl halides and tin amides. Journal of the American Chemical Society, 116(13):5969–5970, 1994....

  47. [55]

    Transfer learning enables the molecular transformer to predict regio- and stereoselective reactions on carbohydrates.Nature communications, 11(1):4874, 2020

    Giorgio Pesciullesi, Philippe Schwaller, Teodoro Laino, and Jean-Louis Reymond. Transfer learning enables the molecular transformer to predict regio- and stereoselective reactions on carbohydrates.Nature communications, 11(1):4874, 2020. doi:10.1038/s41467-020-18671-7

  48. [56]

    Elements of Causal Inference

    Jonas Peters, Dominik Janzing, and Bernhard Schölkopf. Elements of Causal Inference. The MIT Press, 2017

  49. [57]

    MIT Press, 2008

    Joaquin Quiñonero-Candela, Masashi Sugiyama, Anton Schwaighofer, and Neil D Lawrence, editors.Dataset Shift in Machine Learning. MIT Press, 2008

  50. [58]

    RDKit: Open-source cheminformatics, 2021

    RDKit Team. RDKit: Open-source cheminformatics, 2021. URL http://www.rdkit.org

  51. [59]

    The medicinal chemist’ s toolbox: An analysis of reactions used in the pursuit of drug candidates

    Stephen D Roughley and Allan M Jordan. The medicinal chemist’ s toolbox: An analysis of reactions used in the pursuit of drug candidates. Journal of medicinal chemistry, 54(10):3451–3479, 2011. doi:10.1021/jm200187y

  52. [60]

    RXNO: reaction ontologies, 2012

    RSC. RXNO: reaction ontologies, 2012. URL https://github.com/rsc-ontologies/rxno. 14

  53. [61]

    Molecule edit graph attention network: Modeling chemical reactions as sequences of graph edits

    Mikołaj Sacha, Mikołaj Bła˙z, Piotr Byrski, Paweł D ˛ abrowski-Tuma ´ nski, Mikołaj Chromi ´ nski, Rafał Loska, Paweł Włodarczyk-Pruszy ´ nski, and Stanisław Jastrz˛ ebski. Molecule edit graph attention network: Modeling chemical reactions as sequences of graph edits. Journal ...

  54. [62]

    BREEDS: Benchmarks for subpopulation shift

    Shibani Santurkar, Dimitris Tsipras, and Aleksander Madry. BREEDS: Benchmarks for subpopulation shift. In International Conference on Learning Representations 2021, 2021

  55. [63]

    Big data from pharmaceutical patents: A computational analysis of medicinal chemists’ bread and butter.Journal of medicinal chemistry, 59(9):4385–4402, 2016

    Nadine Schneider, Daniel M Lowe, Roger A Sayle, Michael A Tarselli, and Gregory A Landrum. Big data from pharmaceutical patents: A computational analysis of medicinal chemists’ bread and butter.Journal of medicinal chemistry, 59(9):4385–4402, 2016. doi:10.1021/acs.jmedchem.6b00153

  56. [64]

    Found in translation

    Philippe Schwaller, Theophile Gaudin, David Lanyi, Costas Bekas, and Teodoro Laino. “Found in translation”: Predicting outcomes of complex organic chemistry reactions using neural sequence-to-sequence models. Chemical science, 9(28):6091–6098, 2018. doi:10.1039/C8SC02339E

  57. [65]

    Molecular transformer: A model for uncertainty-calibrated chemical reaction prediction

    Philippe Schwaller, Teodoro Laino, Théophile Gaudin, Peter Bolgar, Christopher A Hunter, Costas Bekas, and Alpha A Lee. Molecular transformer: A model for uncertainty-calibrated chemical reaction prediction. ACS Central Science, 5(9):1572–1583, 2019. doi:10.1021/acscentsci.9b00576

  58. [66]

    Predicting retrosynthetic pathways using transformer- based models and a hyper-graph exploration strategy

    Philippe Schwaller, Riccardo Petraglia, Valerio Zullo, Vishnu H Nair, Rico Andreas Haeuselmann, Riccardo Pisoni, Costas Bekas, Anna Iuliano, and Teodoro Laino. Predicting retrosynthetic pathways using transformer- based models and a hyper-graph exploration strategy. Chemical s...

  59. [67]

    On causal and anticausal learning

    Bernhard Schölkopf, Dominik Janzing, Jonas Peters, Eleni Sgouritsa, Kun Zhang, and Joris Mooij. On causal and anticausal learning. In Proceedings of the 29th International Conference on Machine Learning, pages 1255–1262. Omnipress, 2012

  60. [68]

    Modelling chemical reasoning to predict and invent reactions.Chemistry, 23(25):6118–6128, 2017

    Marwin H S Segler and Mark P Waller. Modelling chemical reasoning to predict and invent reactions.Chemistry, 23(25):6118–6128, 2017. doi:10.1002/chem.201604556

  61. [69]

    Planning chemical syntheses with deep neural networks and symbolic AI

    Marwin H S Segler, Mike Preuss, and Mark P Waller. Planning chemical syntheses with deep neural networks and symbolic AI. Nature, 555(7698):604–610, 2018. doi:10.1038/nature25978

  62. [70]

    Improving few- and zero-shot reaction template prediction using modern Hopfield networks

    Philipp Seidl, Philipp Renz, Natalia Dyubankova, Paulo Neves, Jonas Verhoeven, Jörg K Wegner, Marwin Segler, Sepp Hochreiter, and Günter Klambauer. Improving few- and zero-shot reaction template prediction using modern Hopfield networks. Journal of chemical information and mod...

  63. [71]

    Time-split cross-validation as a method for estimating the goodness of prospective predic- tion

    Robert P Sheridan. Time-split cross-validation as a method for estimating the goodness of prospective predic- tion. Journal of chemical information and modeling, 53(4):783–790, 2013. doi:10.1021/ci400084k

  64. [72]

    Improving predictive inference under covariate shift by weighting the log-likelihood function

    Hidetoshi Shimodaira. Improving predictive inference under covariate shift by weighting the log-likelihood function. Journal of statistical planning and inference, 90(2):227–244, 2000. doi:10.1016/s0378-3758(00)00115-4

  65. [73]

    Beyond the imitation game: Quantifying and extrapolating the capabilities of language models

    Aarohi Srivastava and others. Beyond the imitation game: Quantifying and extrapolating the capabilities of language models. Transactions on Machine Learning Research, 2023

  66. [74]

    Lo-Hi: Practical ML drug discovery benchmark

    Simon Steshin. Lo-Hi: Practical ML drug discovery benchmark. In Thirty-seventh Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2023

  67. [75]

    Reproducing the invention of a named reaction: zero-shot prediction of unseen chemical reactions

    An Su, Xinqiao Wang, Ling Wang, Chengyun Zhang, Yejian Wu, Xinyi Wu, Qingjie Zhao, and Hongliang Duan. Reproducing the invention of a named reaction: zero-shot prediction of unseen chemical reactions. Physical chemistry chemical physics: PCCP, 24(17):10280–10291, 2022. doi:10....

  68. [76]

    Machine Learning in Non-Stationary Environments: Introduction to Covariate Shift Adaptation

    Masashi Sugiyama and Kawanabe Motoaki. Machine Learning in Non-Stationary Environments: Introduction to Covariate Shift Adaptation. MIT press, 2012. 15

  69. [77]

    Generative AI for designing and validating easily synthesizable and structurally novel antibiotics

    Kyle Swanson, Gary Liu, Denise B Catacutan, Autumn Arnold, James Zou, and Jonathan M Stokes. Generative AI for designing and validating easily synthesizable and structurally novel antibiotics. Nature machine intelligence, 6(3):338–353, 2024. doi:10.1038/s42256-024-00809-7

  70. [78]

    Intriguing properties of neural networks

    Christian Szegedy, Wojciech Zaremba, Ilya Sutskever, Joan Bruna, Dumitru Erhan, Ian Goodfellow, and Rob Fergus. Intriguing properties of neural networks. arXiv [cs.CV], 2013. doi:10.48550/arXiv.1312.6199

  71. [79]

    State-of-the-art augmented NLP transformer models for direct and single-step retrosynthesis

    Igor V Tetko, Pavel Karpov, Ruud Van Deursen, and Guillaume Godin. State-of-the-art augmented NLP transformer models for direct and single-step retrosynthesis. Nature communications, 11(1):5575, 2020. doi:10.1038/s41467-020-19266-y

  72. [80]

    Unas- sisted noise reduction of chemical reaction data sets

    Alessandra Toniato, Philippe Schwaller, Antonio Cardinale, Joppe Geluykens, and Teodoro Laino. Unas- sisted noise reduction of chemical reaction data sets. Nature Machine Intelligence , 3(6):485–494, 2021. doi:10.1038/s42256-021-00319-w

  73. [81]

    Fast customization of chemical language models to out-of-distribution data sets

    Alessandra Toniato, Alain C Vaucher, Marzena Maria Lehmann, Torsten Luksch, Philippe Schwaller, Marco Stenta, and Teodoro Laino. Fast customization of chemical language models to out-of-distribution data sets. Chemistry of materials: a publication of the American Chemical Soci...

  74. [82]

    Real-world molecular out-of-distribution: Specification and investigation

    Prudencio Tossou, Cas Wognum, Michael Craig, Hadrien Mary, and Emmanuel Noutahi. Real-world molecular out-of-distribution: Specification and investigation. Journal of chemical information and modeling , 2024. doi:10.1021/acs.jcim.3c01774

  75. [83]

    Permutation Invariant Graph-to-Sequence Model for Template-Free Ret- rosynthesis and Reaction Prediction

    Zhengkai Tu and Connor W Coley. Permutation Invariant Graph-to-Sequence Model for Template-Free Ret- rosynthesis and Reaction Prediction. Journal of chemical information and modeling, 62(15):3503–3513, 2022. doi:10.1021/acs.jcim.2c00321

  76. [84]

    Predictive chemistry: machine learning for reac- tion deployment, reaction development, and reaction discovery

    Zhengkai Tu, Thijs Stuyver, and Connor W Coley. Predictive chemistry: machine learning for reac- tion deployment, reaction development, and reaction discovery. Chemical science , 14(2):226–244, 2023. doi:10.1039/d2sc05089g

  77. [85]

    Attention is all you need

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Lukasz Kaiser, and Illia Polosukhin. Attention is all you need. In Advances in Neural Information Processing Systems 30, pages 5998–6008. Curran Associates, Inc., 2017

  78. [86]

    SYNOPSIS: SYNthesize and OPtimize system in silico

    H Maarten Vinkers, Marc R de Jonge, Frederik F D Daeyaert, Jan Heeres, Lucien M H Koymans, Joop H van Lenthe, Paul J Lewi, Henk Timmerman, Koen Van Aken, and Paul A J Janssen. SYNOPSIS: SYNthesize and OPtimize system in silico. Journal of medicinal chemistry, 46(13):2765–2773,...

  79. [87]

    ChemistGA: A chemical synthesizable accessible molecular generation algorithm for real-world drug discovery

    Jike Wang, Xiaorui Wang, Huiyong Sun, Mingyang Wang, Yundian Zeng, Dejun Jiang, Zhenxing Wu, Zeyi Liu, Ben Liao, Xiaojun Yao, Chang-Yu Hsieh, Dongsheng Cao, Xi Chen, and Tingjun Hou. ChemistGA: A chemical synthesizable accessible molecular generation algorithm for real-world d...

  80. [88]

    Heck reaction prediction using a transformer model based on a transfer learning strategy

    Ling Wang, Chengyun Zhang, Renren Bai, Jianjun Li, and Hongliang Duan. Heck reaction prediction using a transformer model based on a transfer learning strategy. Chemical communications (Cambridge, England), 56 (65):9368–9371, 2020. doi:10.1039/d0cc02657c

  81. [89]

    From theory to experiment: transformer-based generation enables rapid discovery of novel reactions

    Xinqiao Wang, Chuansheng Yao, Yun Zhang, Jiahui Yu, Haoran Qiao, Chengyun Zhang, Yejian Wu, Renren Bai, and Hongliang Duan. From theory to experiment: transformer-based generation enables rapid discovery of novel reactions. Journal of cheminformatics, 14(1):60, 2022. doi:10.11...

  82. [90]

    A short review of chemical reaction database systems, computer-aided synthesis design, reaction prediction and synthetic feasibility

    Wendy A Warr. A short review of chemical reaction database systems, computer-aided synthesis design, reaction prediction and synthetic feasibility. Molecular informatics, 33(6-7):469–476, 2014. doi:10.1002/minf.201400052. 16

  83. [91]

    Neural networks for the prediction of organic chemistry reactions

    Jennifer N Wei, David Duvenaud, and Alán Aspuru-Guzik. Neural networks for the prediction of organic chemistry reactions. ACS Central Science, 2(10):725–732, 2016. doi:10.1021/acscentsci.6b00219

  84. [92]

    ORDerly: Data sets and benchmarks for chemical reaction data

    Daniel S Wigh, Joe Arrowsmith, Alexander Pomberger, Kobi C Felton, and Alexei A Lapkin. ORDerly: Data sets and benchmarks for chemical reaction data. Journal of chemical information and modeling, 64(9):3790–3798,

  85. [93]

    HuggingFace’ s transformers: State-of-the-art natural language processing.arXiv [cs.CL],

    Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Rémi Louf, Morgan Funtowicz, Joe Davison, Sam Shleifer, Patrick von Platen, Clara Ma, Yacine Jernite, Julien Plu, Canwen Xu, Teven Le Scao, Sylvain Gugger, Mari...

  86. [94]

    MoleculeNet: a benchmark for molecular machine learning

    Zhenqin Wu, Bharath Ramsundar, Evan N Feinberg, Joseph Gomes, Caleb Geniesse, Aneesh S Pappu, Karl Leswing, and Vijay Pande. MoleculeNet: a benchmark for molecular machine learning. Chemical science, 9(2): 513–530, 2018. doi:10.1039/c7sc02664a

  87. [95]

    OpenOOD: Benchmarking generalized out-of-distribution detection

    Jingkang Yang, Pengyun Wang, Dejian Zou, Zitang Zhou, Kunyuan Ding, Wenxuan Peng, Haoqi Wang, Guangyao Chen, Bo Li, Yiyou Sun, Xuefeng Du, Kaiyang Zhou, Wayne Zhang, Dan Hendrycks, Yixuan Li, and Ziwei Liu. OpenOOD: Benchmarking generalized out-of-distribution detection. In Th...

  88. [96]

    RetroOOD: Understanding out-of-distribution generalization in retrosynthesis prediction

    Yemin Yu, Luotian Yuan, Ying Wei, Hanyu Gao, Xinhai Ye, Zhihua Wang, and Fei Wu. RetroOOD: Understanding out-of-distribution generalization in retrosynthesis prediction. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 38, pages 374–382, 2024. doi:10.16...

  89. [97]

    Retrosynthesis prediction using an end-to-end graph gen- erative architecture for molecular graph editing

    Weihe Zhong, Ziduo Yang, and Calvin Yu-Chian Chen. Retrosynthesis prediction using an end-to-end graph gen- erative architecture for molecular graph editing. Nature communications, 14(1):3009, 2023. doi:10.1038/s41467- 023-38851-5. 17 Appendices S1 Methods In this section we d...

  90. [103]

    the reactants have fewer than 5 heavy atoms,

  91. [104]

    the reactants contain no carbon atoms,

  92. [105]

    none of the reactants have at least two bonds,

  93. [106]

    the reaction is easily identifiable as a (de)protonation (we neutralize commonly occurring charged atoms in the reactants and products using a SMARTS pattern and, having done so, see if the sets of reactants and products are then equal),

  94. [107]

    all products not already present in the reactant set contain fewer than 2 heavy atoms,

  95. [108]

    author-document-dense

    the reaction is very long when tokenized 2 for our language model–based reaction predictor (e.g., over 800 tokens long). 1See https://docs.chemaxon.com/display/docs/formats_chemaxon-extended-smiles-and-smarts-cxsmiles-and-cxsmarts. md 2Details on the tokenizer we use can be fo...

  96. [2012]

    doi:10.1021/ci3003039

  97. [2019]

    doi:10.48550/arXiv.1910.03771

  98. [2020]

    doi:10.48550/arXiv.2010.09885

  99. [2021]

    URL https://www.nextmovesoftware.com/talks/ACS_2021_Fall_Lagerstedt_Namerxn.pdf

  100. [2024]

    doi:10.1021/acs.jcim.4c00292

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.