Pith. sign in

REVIEW 3 major objections 5 minor 42 references

Tango*: Constrained synthesis planning using chemically informed value functions

T0 review · 3 major / 5 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A computed chemical similarity function, TANGO, added to Retro* solves more starting-material-constrained synthesis planning problems than neural-guided baselines, using fewer expansions and less wall-clock time.

desk verdict A simple similarity heuristic beats a learned distance network for constrained retrosynthesis—plausible and worth a referee, but the numbers are partly borrowed from another paper. read the letter →

arxiv 2412.03424 v1 pith:JPFGUIXR submitted 2024-12-04 cs.CE cs.AI

Daniel Armstrong , Zlatko Joncev , Jeff Guo , Philippe Schwaller This is my paper · ORCID
classification cs.CEcs.AI
keywords retrosynthesiscomputer-aidedsynthesisplanningstartingmaterialconstrainedTANGOTanimotosimilarityFuzzyMatchingSubstructureRetro*DESP
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

The paper claims that starting-material-constrained retrosynthesis—planning a route that must end at a specified building block—does not need a specialized bidirectional algorithm or a learned distance network. The authors add a computed node cost, TANGO (a weighted mix of Tanimoto similarity and fuzzy substructure matching to the target starting material), to the existing Retro* search, and show it solves more benchmark (target, starting material) pairs than the neural-network-guided Retro*+D and the bidirectional DESP baselines, with fewer expansions and lower wall-clock time. The same cost function also improves DESP when swapped in for its learned pairwise-distance network, with the best variant reaching a 99.3% solve rate on one benchmark. The paper's diagnostic plots offer a reason: the TANGO cost decreases monotonically and granularly along ground-truth routes, while the learned distance estimates are noisy and often plateau. If true, constrained planning becomes a cheap add-on to general-purpose planners rather than a separate algorithmic family.

What carries the argument

TANGO (TANimoto Group Overlap) is the load-bearing object: a node cost function that measures how structurally similar a molecule node is to any enforced starting material, combining Tanimoto similarity and Fuzzy Matching Substructure (FMS) with a weight $c$, scaled by $k$, and added to the Retro* cost. It is computed directly from molecular structure rather than learned, so it needs no training and no special architecture. Plugged into Retro*'s best-first expansion, it biases the search toward nodes that resemble the target starting material; plugged into DESP in place of its learned pairwise distance network, it yields Tango-F2E and Tango-F2F. The argument that it works rests on the empirical monotonicity of TANGO along ground-truth routes compared with the noisy, plateauing estimates of the learned distance network.

What would settle it

Re-run the three benchmark suites with identical expansion budgets, model checkpoints, and building-block sets, re-running the baselines in the same codebase; if Retro*+D or DESP-F2F matches or beats Tango* under those controlled conditions, the claimed advantage is an artifact of the comparison. Alternatively, retrain the DESP distance network with balanced negative sampling; if its estimates become monotonic and Tango*'s advantage disappears, the paper's mechanism explanation is wrong.

Watch

Extended reading notes

Core claim

The paper's central claim is that the TANGO node cost function—$k\cdot(1-\max_{sm}[c\cdot \mathrm{FMS}(node,sm)+(1-c)\cdot \mathrm{Tanimoto}(node,sm)])$ plus the Retro* cost—guides constrained retrosynthesis better than learned neural guidance. Empirically, Tango(1,0)* outperforms Retro*+D on all three benchmarks (USPTO-190, Pistachio Reachable, Pistachio Hard) at every expansion limit, with solve rates up to 42.6% on USPTO-190 at 500 expansions and 97.3% on Pistachio Reachable, while using fewer expansions and less wall-clock time. Replacing the DESP learned distance network with TANGO yields Tango-F2F, which reaches 99.3% on Pistachio Reachable and roughly 25% higher solve rates than the next-best DESP on the harder datasets. The authors argue the advantage comes from TANGO being a computed, chemically informed value: it is monotonic and granular along true synthetic routes, whereas the neural distance estimates are noisy, plateau, and consistently overestimate synthetic distance.

Load-bearing premise

The comparison assumes the published baseline numbers for Retro*, GRASP, Retro*+D, and DESP were produced under conditions equal to this paper's runs—same expansion budgets, same single-step model and value network checkpoints, same building-block set, and comparable hardware—so the reported solve-rate and efficiency margins could shrink if those conditions differ.

Editorial extensions

If this is right

  • Constrained synthesis planning can be achieved by adding a computed similarity term to an existing uni-directional planner like Retro*, with no model retraining.
  • TANGO can replace learned pairwise-distance networks in bidirectional planners; Tango-F2F becomes the strongest solver in the comparison.
  • Because searches guided by TANGO expand fewer nodes, wall-clock time stays low even though the similarity computation adds per-node overhead.
  • The same cost idea can be pointed at other structural goals—key intermediates or substructures—not just full starting materials.
  • Route lengths are comparable to or shorter than neural-guided baselines, so the efficiency gain does not come at the cost of longer routes.

Reading between the lines

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

  • If TANGO's advantage comes from its monotonicity rather than chemical specificity, then other cheap, well-calibrated structural heuristics (e.g., learned embeddings trained with a ranking objective) might reproduce the gain; the paper does not test this.
  • The diagnostic failure of the learned distance network suggests its negative-sample training (pairs assigned a fixed distance of 10) may be the culprit; rebalancing those samples could close the gap between learned and computed guidance.
  • Because TANGO is computed from structure, it should transfer across reaction datasets without re-tuning; the paper only shows transfer of hyperparameters from one benchmark to harder sets, not across different single-step models.
  • The current constraint is a single molecule at the goal; extending the max over starting materials already works, but enforcing multiple structural constraints along one route would require a different composition rule.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper introduces Tango*, a node-cost function for starting-material-constrained retrosynthetic planning. TANGO combines Tanimoto similarity and Fuzzy Matching Substructure between a molecule and the specified starting material; the cost is k*(1 - max_sm [c*Tanimoto + (1-c)*FMS]) added to the Retro* cost. The authors integrate this cost into Retro* and into the bidirectional DESP-F2E/F2F algorithms, and evaluate on USPTO-190, Pistachio Reachable, and Pistachio Hard. They report that Tango(1,0)* outperforms the neural-network-guided Retro*+D and matches or exceeds DESP baselines in solve rate, with fewer expansions and lower wall-clock time. They also analyze value-function monotonicity along ground-truth routes and present case studies, including a route to chlorambucil from renewable feedstocks.

Significance. If the reported results are reproducible under controlled comparison, the contribution is notable: a simple, non-learned similarity heuristic can replace a learned pairwise distance network for constrained synthesis planning, improving both efficiency and solve rate. The paper is transparent about using the same single-step and value-network checkpoints as Yu et al. and discloses code. The hyperparameter tuning on one dataset and transfer to others is a strength, as are the literature-validated case-study routes. However, the headline comparisons against Retro*, GRASP, Retro*+D, and DESP rely on numbers imported from a different paper, which currently prevents the efficiency and solve-rate claims from being considered fully controlled.

major comments (3)
  1. [Section 4.1, Tables 1 and 2] All baseline numbers for Retro*, GRASP, Retro*+D, and DESP-F2E/F2F are taken from Yu et al. rather than re-run under the same conditions. The paper's central claims of 'consistently outperforms' and 'lower wall clock times' depend on equivalence of expansion-budget semantics (model calls vs node expansions), single-step model and value-network checkpoints, the building-block set, and hardware. The authors use Yu et al.'s checkpoints and an implementation based on the DESP codebase, which makes equivalence plausible, but it is not demonstrated. This is load-bearing because every headline result is a comparison to these external numbers; the authors' own ablations only compare Tango* variants to each other. I ask that the baselines be re-run under the authors' exact evaluation harness, or, failing that, that the paper provide a detailed point-by-point argument for equivalence of all four factors listed above.
  2. [Section 4.1 and Algorithm 1] The definition of the FMS/Tanimoto weighting parameter c is internally inconsistent. Algorithm 1 computes reward_sm = TanSim·c + FMS·(1−c), which makes c the Tanimoto weight. The text, however, states that 'c defines the FMS weight' and then refers to 'Tango with c = 0.0 as Tango(1, 0)'. Under the formula in Algorithm 1, c=0 gives reward_sm = FMS, i.e., pure FMS guidance, not pure Tanimoto as the name Tango(1,0) implies. This makes the exact cost function ambiguous and harms reproducibility. Please correct either the algorithm or the naming convention.
  3. [Abstract and Section 4.1] The abstract claims that 'by optimising a single hyperparameter, Tango* outperforms existing methods', but the method in fact tunes two hyperparameters, k and c, and the paper's main results use Tango(1,0) with c=0.0, which is not the value found optimal by the hyperparameter screen (c=0.3, i.e., Tango(0.7,0.3), improves Pistachio Reachable but not the harder datasets). The authors should clarify whether the proposed method is the tuned configuration or the manually selected c=0.0 configuration, and should state the tuning procedure accurately. This matters because the abstract's 'single hyperparameter' claim and the choice of the headline configuration are currently presented inconsistently.
minor comments (5)
  1. [Section 4.2] The text uses 'UPSTO-190' where 'USPTO-190' is meant.
  2. [Figure 1 caption] The word 'constrainted' should be 'constrained'.
  3. [Appendix A.1] The CPU model is listed as 'AMD Rysen 9 7900X'; the correct spelling is 'Ryzen'.
  4. [Section 4.4] The paragraph ending with 'perform substantially better at estimating synthetic distance. unprivileged setting.' contains an incomplete final sentence fragment; please rewrite.
  5. [Table 2] The caption states 'Route length comparisons are made on the routes solved by all methods', but it is not specified whether the wall-clock times are also computed only on the common subset or over the entire benchmark; please clarify.

Circularity Check

1 steps flagged · score 4.0 of 10

Partial circularity: Pistachio Reachable benchmark is also the tuning set for the Tango* hyperparameter k, so its reported solve rates are in-sample, while other benchmarks remain independent.

  1. fitted input called prediction [Section 4.1 (Hyperparameter Optimisation) and Table 1]
    "To evaluate the ability of our method to generalise from simpler to more complex molecules, we choose the Pistachio Reachable dataset for hyperparameter tuning. We find a value of k = 25 optimises both Solve Rate and Average Number of expansions."

    The same dataset used to select the Tango* weight k is then reported as a benchmark in Table 1, so the Pistachio Reachable solve rates for Tango(1,0)* are the result of fitting rather than independent prediction. For example, the reported solve rate at expansion budget 50 was directly used to select k=25, making that column in-sample. The USPTO-190 and Pistachio Hard results are not affected by this particular fit, and the main Tango(1,0)* configuration uses c=0.0 rather than the optimized c=0.3, so the circularity is limited to one of three benchmarks. Still, the abstract's claim that Tango* outperforms 'in terms of efficiency and solve rate' partially rests on a value that was tuned on the same benchmark.

full rationale

The TANGO cost function is a fixed, computed similarity-based heuristic with no fitted coefficients beyond the disclosed hyperparameters k and c. The central derivation that TANGO guides search toward the starting material is self-contained: it is a direct additive combination of Tanimoto similarity and fuzzy substructure overlap with the goal molecule, and the paper does not train it on the benchmark data. The monotonicity analysis in Section 4.4 uses ground truth routes as an external reference, so it is not circular fitting. The only identifiable circular step is the hyperparameter tuning: k is selected on Pistachio Reachable and then the same dataset is reported as a benchmark. This makes the Pistachio Reachable column in-sample for Tango(1,0)*, though the other two benchmarks (USPTO-190 and Pistachio Hard) are unaffected. The comparison to cross-paper baselines from Yu et al. is a fairness/correctness concern, not circularity, because the baselines are external numbers and the paper's own Tango runs are on its own implementation. Overall, the central claim that a simple similarity heuristic can replace a learned distance network retains independent content on the harder benchmarks, so the circularity score is moderate (4) rather than higher.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The paper introduces no new physical or chemical entities. Its contribution is a new cost function built from standard similarity metrics. The only fitted values are the two hyperparameters k and c, both tuned on a single benchmark and held fixed elsewhere. The central claim rests on the validity of molecular similarity as a search signal and on the fairness of the baseline comparison.

free parameters (2)
  • k (TANGO weight) = 25
    Scalar multiplying the TANGO reward in the node cost; tuned on Pistachio Reachable (Section 4.1).
  • c (FMS weight) = 0.3 for Tango(0.7,0.3), 0.0 for Tango(1,0)
    Controls the balance of Tanimoto similarity and fuzzy substructure matching; tuned on Pistachio Reachable. Main results use c=0.0.
assumptions (3)
  • domain assumption Tanimoto similarity and FMS, computed on molecular fingerprints, are meaningful chemical similarity measures for guiding retrosynthesis.
    Invoked in Algorithm 1, the paper's entire guidance signal. The paper does not validate which fingerprint or substructure parameters are used.
  • domain assumption A monotonic decrease of node cost along ground truth synthetic routes improves search performance.
    Section 4.4, Eq. 1, used to explain Tango*'s advantage. This is a hypothesis, not proven.
  • domain assumption The single-step retrosynthesis model and Retro* value network from Yu et al. are used as fixed black boxes of sufficient quality.
    Section 4.1: 'we use the Retro* value network and single-step retrosynthesis model provided by Yu et al.' The comparison assumes these models are appropriate for all methods.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Tango*: Constrained synthesis planning using chemically informed value functions." pith.science (2026). https://pith.science/paper/JPFGUIXR

@misc{pith2026241203424,
  author       = {Pith},
  title        = {Pith review of: Tango*: Constrained synthesis planning using chemically informed value functions},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JPFGUIXR}},
  note         = {Machine review of arXiv:2412.03424}
}
read the original abstract

Computer-aided synthesis planning (CASP) has made significant strides in generating retrosynthetic pathways for simple molecules in a non-constrained fashion. Recent work introduces a specialised bidirectional search algorithm with forward and retro expansion to address the starting material-constrained synthesis problem, allowing CASP systems to provide synthesis pathways from specified starting materials, such as waste products or renewable feed-stocks. In this work, we introduce a simple guided search which allows solving the starting material-constrained synthesis planning problem using an existing, uni-directional search algorithm, Retro*. We show that by optimising a single hyperparameter, Tango* outperforms existing methods in terms of efficiency and solve rate. We find the Tango* cost function catalyses strong improvements for the bidirectional DESP methods. Our method also achieves lower wall clock times while proposing synthetic routes of similar length, a common metric for route quality. Finally, we highlight potential reasons for the strong performance of Tango over neural guided search methods

Figures

Figures reproduced from arXiv: 2412.03424 by the authors.

Figure 1
Figure 1. Comparison of existing constrained synthesis planning methods with Tango* [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Here we demonstrate a meaningful 12-step route generated by our method on a [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. A comparison of node cost estimates for USPTO-190 routes solved and not solved by [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Here we demonstrate a feasible 10-step route generated by Tango-DESP-F2F on a [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Here we show a feasible synthesis route to the chemotherapy drug, Chlorambucil, a WHO [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: A hyper-parameter screen conducted on Pistachio Reachable with an expansion limit of 50. [PITH_FULL_IMAGE:figures/full_fig_p014_6.png]
Figure 7
Figure 7. Figure 7: Failure points of value networks (Retro* and SynDist) compared with ground truth synthetic [PITH_FULL_IMAGE:figures/full_fig_p015_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

42 extracted references · 36 canonical work pages

  1. [1]

    General methods for the construction of complex molecules

    Elias James Corey. General methods for the construction of complex molecules. Pure and Applied chemistry, 14:19–38, 1967

    work page 1967

  2. [2]

    E. J. Corey and Xue-Min Chelg. The Logic of Chemical Synthesis. John Wiley & Sons, 1995

    work page 1995

  3. [3]

    Computer-assisted design of complex organic syntheses: Pathways for molecular synthesis can be devised with a computer and equipment for graphical communication

    Elias James Corey and W Todd Wipke. Computer-assisted design of complex organic syntheses: Pathways for molecular synthesis can be devised with a computer and equipment for graphical communication. Science, 166(3902):178–192, 1969

    work page 1969

  4. [4]

    Chemical reactions from US patents (1976-Sep2016) http://doi.org/10

    Daniel Lowe. Chemical reactions from US patents (1976-Sep2016) http://doi.org/10. 6084/m9.figshare.5104873.v1, 6 2017

    work page 1976

  5. [5]

    Retrosynthetic re- action prediction using neural sequence-to-sequence models

    Bowen Liu, Bharath Ramsundar, Prasad Kawthekar, Jade Shi, Joseph Gomes, Quang Luu Nguyen, Stephen Ho, Jack Sloane, Paul Wender, and Vijay Pande. Retrosynthetic re- action prediction using neural sequence-to-sequence models. ACS Cent. Sci., 3:1103–1113, 2017

    work page 2017

  6. [6]

    Neural-symbolic machine learning for retrosynthesis and reaction prediction

    Marwin HS Segler and Mark P Waller. Neural-symbolic machine learning for retrosynthesis and reaction prediction. Chem. Eur. J., 23:5966–5971, 2017

    work page 2017

  7. [7]

    Computer-assisted retrosynthesis based on molecular similarity

    Connor W Coley, Luke Rogers, William H Green, and Klavs F Jensen. Computer-assisted retrosynthesis based on molecular similarity. ACS Cent. Sci., 3:1237–1245, 2017

    work page 2017

  8. [8]

    Retro*: Learning retrosynthetic planning with neural guided A* search

    Binghong Chen, Chengtao Li, Hanjun Dai, and Le Song. Retro*: Learning retrosynthetic planning with neural guided A* search. In International Conference on Machine Learning, pages 1608–1616. PMLR, 2020

    work page 2020

Show all 42 references
  1. [9]

    Permutation invariant graph-to-sequence model for template- free retrosynthesis and reaction prediction

    Zhengkai Tu and Connor W Coley. Permutation invariant graph-to-sequence model for template- free retrosynthesis and reaction prediction. arXiv preprint arXiv:2110.09681, 2021

    2021 arXiv

  2. [10]

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

    Mikolaj Sacha, Mikolaj Blaz, Piotr Byrski, Pawel Dabrowski-Tumanski, Mikolaj Chrominski, Rafal Loska, Pawel Wlodarczyk-Pruszynski, and Stanislaw Jastrzebski. Molecule edit graph attention network: Modeling chemical reactions as sequences of graph edits. J. Chem. Inf. Model., 6...

    2021

  3. [11]

    Planning chemical syntheses with deep neural networks and symbolic ai

    Marwin HS Segler, Mike Preuss, and Mark P Waller. Planning chemical syntheses with deep neural networks and symbolic ai. Nature, 555(7698):604–610, 2018

    2018

  4. [12]

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

    Philippe Schwaller, Riccardo Petraglia, Valerio Zullo, Vishnu H Nair, Rico Andreas Haeusel- mann, Riccardo Pisoni, Costas Bekas, Anna Iuliano, and Teodoro Laino. Predicting retrosyn- thetic pathways using transformer-based models and a hyper-graph exploration strategy. Chem. S...

    2020

  5. [13]

    Depth-first proof-number search with heuristic edge cost and application to chemical synthesis planning

    Akihiro Kishimoto, Beat Buesser, Bei Chen, and Adi Botea. Depth-first proof-number search with heuristic edge cost and application to chemical synthesis planning

  6. [14]

    AiZynthFinder: a fast, robust and flexible open-source software for retrosynthetic planning

    Samuel Genheden, Amol Thakkar, Veronika Chadimová, Jean-Louis Reymond, Ola Engkvist, and Esben Bjerrum. AiZynthFinder: a fast, robust and flexible open-source software for retrosynthetic planning. J. Cheminf., 12:1–9, 2020

    2020

  7. [15]

    New and efficient approaches to the semisynthesis of taxol and its c-13 side chain analogs by means of β-lactam synthon method

    Iwao Ojima, Ivan Habus, Mangzhu Zhao, Martine Zucco, Young Hoon Park, Chung Ming Sun, and Thierry Brigaud. New and efficient approaches to the semisynthesis of taxol and its c-13 side chain analogs by means of β-lactam synthon method. Tetrahedron, 48(34):6985–7012, 1992

    1992

  8. [16]

    Navigating the chiral pool in the total synthesis of complex terpene natural products

    Zachary G Brill, Matthew L Condakes, Chi P Ting, and Thomas J Maimone. Navigating the chiral pool in the total synthesis of complex terpene natural products. Chemical reviews, 117 (18):11753–11795, 2017

    2017

  9. [17]

    Herrera, Josef M

    Agnieszka Wołos, Dominik Koszelewski, Rafał Roszak, Sara Szymku ´c, Martyna Moskal, Ryszard Ostaszewski, Brenden T. Herrera, Josef M. Maier, Gordon Brezicki, Jonathon Samuel, Justin A. M. Lummiss, D. Tyler McQuade, Luke Rogers, and Bartosz A. Grzybowski. Computer- designed rep...

    2022

  10. [18]

    Application of automated network generation for retrosynthetic planning of potential corrosion inhibitors

    Lauren M Lopez, Quan Zhang, Orion Dollar, Jim Pfaendtner, Brent H Shanks, and Linda J Broadbelt. Application of automated network generation for retrosynthetic planning of potential corrosion inhibitors. Molecular Systems Design & Engineering, 9(4):352–371, 2024

    2024

  11. [19]

    Computational synthesis design for controlled degradation and revalorization

    Anna ˙Z ˛ adło-Dobrowolska, Karol Molga, Olga O Kolodiazhna, Sara Szymku´c, Martyna Moskal, Rafał Roszak, and Bartosz A Grzybowski. Computational synthesis design for controlled degradation and revalorization. Nature Synthesis, pages 1–12, 2024

    2024

  12. [20]

    Starting material oriented retrosynthetic analysis in the lhasa program

    A Peter Johnson, Chris Marshall, and Philip N Judson. Starting material oriented retrosynthetic analysis in the lhasa program. 1. general description. Journal of chemical information and computer sciences, 32(5):411–417, 1992. 10

    1992

  13. [21]

    Grasp: Navigating retrosynthetic planning with goal-driven policy

    Yemin Yu, Ying Wei, Kun Kuang, Zhengxing Huang, Huaxiu Yao, and Fei Wu. Grasp: Navigating retrosynthetic planning with goal-driven policy. Advances in Neural Information Processing Systems, 35:10257–10268, 2022

    2022

  14. [22]

    Double-ended synthesis planning with goal-constrained bidirectional search

    Kevin Yu, Jihye Roh, Ziang Li, Wenhao Gao, Runzhong Wang, and Connor W Coley. Double-ended synthesis planning with goal-constrained bidirectional search. arXiv preprint arXiv:2407.06334, 2024

    2024 arXiv

  15. [23]

    SMILES, a chemical language and information system

    David Weininger. SMILES, a chemical language and information system. 1. introduction to methodology and encoding rules. J. Chem. Inf. Comput. Sci., 28:31–36, 1988

    1988

  16. [24]

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

    Weihe Zhong, Ziduo Yang, and Calvin Yu-Chian Chen. Retrosynthesis prediction using an end-to-end graph generative architecture for molecular graph editing. Nature Communications, 14(1):3009, 2023

    2023

  17. [25]

    G 2 retro as a two-step graph generative models for retrosynthesis prediction

    Ziqi Chen, Oluwatosin R Ayinde, James R Fuchs, Huan Sun, and Xia Ning. G 2 retro as a two-step graph generative models for retrosynthesis prediction. Communications Chemistry, 6 (1):102, 2023

    2023

  18. [26]

    Lhasa—logic and heuristics applied to synthetic analysis

    David A Pensak and Elias James Corey. Lhasa—logic and heuristics applied to synthetic analysis. ACS Publications, 1977

    1977

  19. [27]

    Computer-assisted synthetic planning: the end of the beginning

    Sara Szymku ´c, Ewa P Gajewska, Tomasz Klucznik, Karol Molga, Piotr Dittwald, Michał Startek, Michał Bajczyk, and Bartosz A Grzybowski. Computer-assisted synthetic planning: the end of the beginning. Angew. Chem. - Int. Ed., 55:5904–5937, 2016

    2016

  20. [28]

    Learning retrosynthetic planning through simulated experience

    John S Schreck, Connor W Coley, and Kyle JM Bishop. Learning retrosynthetic planning through simulated experience. ACS central science, 5(6):970–981, 2019

    2019

  21. [29]

    Self-improved retrosynthetic planning

    Junsu Kim, Sungsoo Ahn, Hankook Lee, and Jinwoo Shin. Self-improved retrosynthetic planning. arXiv preprint arXiv:2106.04880, 2021

    2021 arXiv

  22. [30]

    Retrosynthetic planning with experience-guided monte carlo tree search

    Siqi Hong, Hankz Hankui Zhuo, Kebing Jin, Guang Shao, and Zhanwen Zhou. Retrosynthetic planning with experience-guided monte carlo tree search. Communications Chemistry, 6(1): 120, 2023

    2023

  23. [31]

    Retrosynthetic planning with dual value networks

    Guoqing Liu, Di Xue, Shufang Xie, Yingce Xia, Austin Tripp, Krzysztof Maziarz, Marwin Segler, Tao Qin, Zongzhang Zhang, and Tie-Yan Liu. Retrosynthetic planning with dual value networks. In International Conference on Machine Learning , pages 22266–22276. PMLR, 2023

    2023

  24. [32]

    Datasets and their influence on the development of computer assisted synthesis planning tools in the pharmaceutical domain

    Amol Thakkar, Thierry Kogej, Jean-Louis Reymond, Ola Engkvist, and Esben Jannik Bjerrum. Datasets and their influence on the development of computer assisted synthesis planning tools in the pharmaceutical domain. Chem. Sci., 11:154–168, 2020

    2020

  25. [33]

    Re-evaluating retrosynthesis algorithms with syntheseus

    Krzysztof Maziarz, Austin Tripp, Guoqing Liu, Megan Stanley, Shufang Xie, Piotr Gai ´nski, Philipp Seidl, and Marwin Segler. Re-evaluating retrosynthesis algorithms with syntheseus. arXiv preprint arXiv:2310.19796, 2023

    2023 arXiv

  26. [34]

    Paula Torren-Peraire, Alan Kai Hassen, Samuel Genheden, Jonas Verhoeven, Djork-Arné Clevert, Mike Preuss, and Igor V . Tetko. Models matter: the impact of single-step retrosynthesis on synthesis planning. Digital Discovery, 3:558–572, 2024. doi: 10.1039/D3DD00252G. URL http://...

    2024 doi

  27. [35]

    Enhancing diversity in language based models for single-step retrosynthesis

    Alessandra Toniato, Alain C Vaucher, Philippe Schwaller, and Teodoro Laino. Enhancing diversity in language based models for single-step retrosynthesis. Digital Discovery, 2(2): 489–501, 2023

    2023

  28. [36]

    Unbiasing retrosynthesis language models with disconnection prompts

    Amol Thakkar, Alain C Vaucher, Andrea Byekwaso, Philippe Schwaller, Alessandra Toniato, and Teodoro Laino. Unbiasing retrosynthesis language models with disconnection prompts. ACS Central Science, 9(7):1488–1498, 2023

    2023

  29. [37]

    Constrained synthesis planning with disconnection-aware transformer and multi-objective search

    Annie M Westerlund, Lakshidaa Saigiridharan, and Samuel Genheden. Constrained synthesis planning with disconnection-aware transformer and multi-objective search. 2024

    2024

  30. [38]

    It takes two to tango: Directly optimizing for constrained synthesizability in generative molecular design

    Jeff Guo and Philippe Schwaller. It takes two to tango: Directly optimizing for constrained synthesizability in generative molecular design. arXiv preprint arXiv:2410.11527, 2024

    2024 arXiv

  31. [39]

    Gajewska, Sara Szymku´c, Piotr Dittwald, Michał Startek, Oskar Popik, Jacek Mlynarski, and Bartosz A

    Ewa P. Gajewska, Sara Szymku´c, Piotr Dittwald, Michał Startek, Oskar Popik, Jacek Mlynarski, and Bartosz A. Grzybowski. Algorithmic Discovery of Tactical Combinations for Advanced Organic Syntheses. Chem, 6(1):280–293, 2020. ISSN 2451-9294. doi: https://doi.org/10.1016/ j.che...

    2020

  32. [40]

    Combining retrosynthesis and mixed-integer optimization for minimizing the chemical inventory needed to realize a who essential medicines list

    Hanyu Gao, Connor W Coley, Thomas J Struble, Linyan Li, Yujie Qian, William H Green, and Klavs F Jensen. Combining retrosynthesis and mixed-integer optimization for minimizing the chemical inventory needed to realize a who essential medicines list. Reaction Chemistry & Enginee...

    2020

  33. [41]

    Iridium-catalysed reductive deoxygenation of ketones with formic acid as traceless hydride donor

    Zhiheng Yang, Xueya Zhu, Shiyi Yang, Weiyan Cheng, Xiaojian Zhang, and Zhanhui Yang. Iridium-catalysed reductive deoxygenation of ketones with formic acid as traceless hydride donor. Advanced Synthesis & Catalysis, 362(23):5496–5505, 2020

    2020

  34. [42]

    reaction number

    Aleem Gangjee, Ying Wang, Sherry F. Queener, and Roy L. Kisliuk. Synthesis of 2,6-diamino- 5-[(2-substituted phenylamino)ethyl]pyrimidin-4(3h)-one as inhibitors of folate metabolizing enzymes. Journal of Heterocyclic Chemistry, 43, March 2009. doi: 10.1002/jhet.5570430615. Pre...

    2009 doi

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.