Pith. sign in

REVIEW 3 major objections 4 minor 76 references

Strategy-first synthesis planning for complex natural products

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A strategy-first LLM planner designs complete routes to complex natural products without a template library.

desk verdict A well-scoped, honestly reported advance in LLM synthesis planning: the reach claim holds at graph level, chemical validity is unverified but the paper says so plainly. read the letter →

arxiv 2608.07454 v1 pith:IAOKL5B7 submitted 2026-08-07 cs.MA cs.AI

Daniel Armstrong , Xuan-Vu Nguyen , Octavian Susanu , Gabriel Gibberd , Théo A. Neukomm , Taddäus Strunden , Dan Forster , Morgane Delattre
show 10 more authors
Shawn Teh Clément Rols John Federice Hayden Leatherwood M. Lavelle Barnes Maarten R. Dobbelaere Peter Wipf Jon T. Njardarson Jieping Zhu Philippe Schwaller
This is my paper · ORCID
classification cs.MAcs.AI
keywords retrosynthesisnaturalproductsynthesislargelanguagemodelsagenticplanningtemplate-freereactionrepresentationspacetotalmulti-agentsystems
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 a multi-agent LLM planner, SynthEx, can design full retrosynthetic routes to complex natural products that have no published synthesis, by writing each disconnection as an ordered list of atom-level graph edits instead of choosing from a fixed reaction template library. On a benchmark of 1,098 such targets, it returns complete routes to purchasable building blocks for 63.9% of them, while a near-exhaustively run template-based planner solves 13.8%. Blinded expert chemists rated SynthEx's key steps statistically indistinguishable from published human syntheses on feasibility, elegance, and overall quality, with only a small gap in strategic value. The authors release the routes as an open database, SynthAtlas, and treat experimental validation as the next frontier.

What carries the argument

ReactionJSON, a template-free reaction representation in which a retrosynthetic disconnection is an ordered list of atom-level graph-edit operations (break_bond, add_bond, change_bond_order, add/remove group, and stereochemistry operations) applied to an atom-mapped product to yield the precursors deterministically. Because a route is serialized as RouteJSON, a text object anchored on atom maps, it can be edited in place: a Critic agent simulates each reaction forward, flags 'blocking' steps, and an Editor performs surgical repairs such as reordering steps, inserting protections, or changing conditions without restarting the search. This representation is what converts the LLM from a selector among templates into a generator of novel disconnections, and it is what makes the iterative self-correction loop practical.

What would settle it

Select a representative sample of SynthEx's proposed key steps, especially ring-forming and stereoselective steps, and attempt them in the laboratory under the abstract conditions the routes annotate; if a substantial fraction fail or give incorrect stereochemistry, the expert-comparable step quality and the 63.9% solve rate would not reflect experimentally viable planning.

Watch

Extended reading notes

Core claim

The central claim is that a synthesis planner can operate inventively outside the reaction space defined by patent-derived corpora. SynthEx expresses every disconnection as ReactionJSON, an ordered list of atom-level graph edits applied to the atom-mapped product, so the language model generates novel expansions rather than selecting among templates. On 1,098 structurally complex natural products with no reported total synthesis, SynthEx produces complete routes to purchasable building blocks for 63.9% of targets versus 13.8% for a template-based planner run under a generous budget; the advantage widens with molecular complexity, and the proposed chemistry is more convergent and far richer in ring-forming steps (16.0% of steps, against 2.8% for a state-of-the-art corpus-trained single-step model's top-1 predictions). In blinded evaluation, expert chemists rated SynthEx's key steps comparable to published human syntheses on feasibility, elegance, and overall quality, while a small but detectable gap remained on strategic value. The paper is explicit that these are graph-level proposals: stereochemical outcomes were not verified, and the improvement loop is scored by the same class of model that performs the repairs.

Load-bearing premise

The load-bearing premise is that a route judged chemically plausible by an LLM critic and by name-recognition software is a genuine synthesis plan; because no reactions were run and stereochemical outcomes were not verified, a large experimental failure rate among proposed steps would mean the 63.9% solve rate and the expert-comparable ratings overstate what the planner can actually deliver.

Editorial extensions

If this is right

  • If correct, retrosynthetic planning no longer needs to be confined to a fixed reaction library; an LLM that writes graph edits can propose chemistry that is rare in patent corpora but recognizable to expert chemists.
  • The solve-rate advantage grows with target complexity, reaching 56% on the large complex subset where the template baseline solves only 4%, so the method is most useful exactly where conventional planners fail hardest.
  • The released SynthAtlas corpus of 33,145 atom-mapped reactions provides a large, dated set of public predictions for molecules nobody has yet synthesized, enabling future concordance tracking as real syntheses appear.
  • On targets where both planners succeed, SynthEx routes are shorter than template-planner routes (median 5 steps versus 11), suggesting that its strategic disconnections simplify the search problem for downstream completion.
  • The expert-comparable ratings on feasibility, elegance, and overall quality imply that SynthEx's key steps could serve as credible starting points for experimental campaigns, with strategic guidance remaining the one axis where human input still adds value.

Reading between the lines

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

  • If the 63.9% solve rate survives wet-lab testing, the bottleneck in automated synthesis shifts from proposing disconnections to predicting reaction conditions and stereochemical outcomes, which the paper itself flags as unverified.
  • Because SynthAtlas is atom-mapped and lies outside patent-derived corpora, it could become training and evaluation data for next-generation single-step retrosynthesis models, a use the paper mentions but does not itself pursue.
  • The small but detectable strategic-value gap suggests the most productive near-term arrangement is a chemist supplying the high-level strategy while the agent elaborates and critiques the route, a division the paper identifies as promising.
  • A direct testable extension is to apply the planner to targets where literature routes failed at a specific step, such as the Melonine Mannich cyclization, and see whether its alternative disconnections reproducibly avoid the documented failure mode.
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 / 4 minor

Summary. The manuscript introduces SynthEx, a multi-agent LLM retrosynthesis planner in which disconnections are written directly as atom-mapped graph edits (ReactionJSON) rather than selected from a fixed template library. A strategy generator proposes several high-level strategies, a route builder expands them into full routes, a critic/editor loop repairs flagged steps, and an analyst scores feasibility. The system is benchmarked on 1,098 NPAtlas natural products with no reported total syntheses, where it returns leaf-purchasable routes for 63.9% of targets versus 13.8% for a near-exhaustively resourced AiZynthFinder baseline; the strategic layer alone reaches 25.0%. Reaction-space analyses show that SynthEx's steps are poorly recovered by USPTO-derived classifiers and by RetroChimera, and that ring-forming, convergent chemistry is overrepresented relative to patent-derived corpora. In a blinded expert study, ten synthetic chemists rated 148 key steps from 47 targets on which SynthEx's strategy was judged congruent with a published synthesis; SynthEx was statistically indistinguishable from literature steps on feasibility, elegance, and overall quality, with a small but detectable deficit on strategic value. The routes are released as the open SynthAtlas resource.

Significance. The contribution is potentially significant. The ReactionJSON representation is a clean, falsifiable mechanism for stepping outside patent-derived reaction space, and the main reach comparison is benchmarked against an external open-source planner (AiZynthFinder) and external classifiers (NameRXN, RetroChimera), so the central comparison is not circular. The authors are also commendably explicit about what the pipeline does not establish: they state that stereochemical outcomes were not verified, that the critic/editor loop is an internal consistency procedure, and that the expert comparison is conditional on strategic congruence. The release of dated, public route predictions for unsynthesized natural products is a genuine strength with predictive value. The main weakness is that the headline 'solved' criterion is graph-level leaf purchasability, so the quantitative reach claims and the 'comparable to human' conclusion are both contingent on chemical validity that is asserted rather than demonstrated.

major comments (3)
  1. [§4.4, §2.4] The solved criterion in §4.4 counts a target as solved when every leaf InChIKey is in the ZINC/eMolecules stock, with no check that the ReactionJSON graph edits correspond to chemically viable transformations with correct regio- and stereochemistry. The paper's own Discussion concedes that stereochemical outcomes were not verified and that expert review surfaced occasional selectivity and feasibility errors. Because the 63.9% headline and the 'beyond the reach of conventional planners' claim rest on this criterion, the metric should be relabeled as a graph-completion rate or supplemented with a forward-chemistry validation (e.g., a trained forward model or a blinded expert audit of complete routes, not just key steps) so that the reader can estimate what fraction of proposed steps are plausibly executable. As written, the abstract's phrase 'plans routes' overstates what the evaluation demonstrates.
  2. [§4.10, Appendix B] The expert comparison is conditioned on 47 of 70 targets retained after an LLM judged SynthEx's strategy 'congruent' with the published synthesis, and the manuscript provides no reliability assessment of that congruence judgment. Since the same model family that generates the routes also selects the comparison set, the retained targets may be systematically easier or more similar to literature chemistry, and the expert ratings cannot speak to the full benchmark. In addition, the literature key steps are nominated by a different LLM and the SynthEx key steps are the tool's own nominations; the comparison therefore measures steps each source regards as pivotal, not a matched random sample of steps. I ask for a sensitivity analysis on all 70 targets, or a human audit of congruence on a random subset, and for the paper to state explicitly how the 47/70 selection affects generalization of the 'comparable to human' conclusion.
  3. [§4.3, §2.5] The route-improvement evidence in Fig. 5 is generated and scored by agents sharing the same LLM backbone: the Critic labels blocking reactions, the Editor repairs them, and the Analyst re-scores feasibility. As the authors note, this is an internal consistency procedure rather than external validation, but the figure and the surrounding text present the declining blocking rate and the feasibility shift as evidence of quality improvement. Because shared blind spots could make the loop converge to a self-consistent but chemically invalid state, the improvement claim should either be validated by an independent mechanism (a different model family, a forward-reaction predictor, or human review of a random sample of before/after routes) or be explicitly reframed as measuring only convergence to the Critic's own criterion.
minor comments (4)
  1. [§4.1, §2.4] The control subset is constructed from targets that initially failed a short template search, so the 80% solve rate under a generous budget cleanly demonstrates a budget-limited failure; however, the subset is small (n=123) and the manuscript does not report confidence intervals for the per-subset solve rates, which would help the reader judge the strength of the comparison.
  2. [Fig. 4f] The target-wise comparison of route lengths between SynthEx and the literature is not apples-to-apples, since literature routes are experimentally optimized and SynthEx routes are untested first proposals; the paper makes this point in the text, but the figure panel as displayed may still invite an unfavorable comparison that the SI D.2 joint-solved comparison addresses more fairly.
  3. [Abstract, §2.4] The abstract states that expert chemists 'engaged with them as genuine synthesis plans,' which is stronger than the evidence: the raters saw individual key steps, not complete routes, and only on the 47 strategically congruent targets. Please align the abstract's wording with the actual scope of the expert study.
  4. [§4.2, §2.2.1] The knowledge-cutoff argument is appropriately cautious, but the claim that SynthEx 'recovers' the Okaramine M route 'after the training cut-off' rests on API documentation rather than on a verified absence from post-training data; the paper's own caveat that post-training data is not disclosed should perhaps appear in the main-text discussion of this case, not only in the Methods.

Circularity Check

1 steps flagged · score 2.0 of 10

Central reach and expert-comparison claims are benchmarked externally and are not circular; only the explicitly disclosed self-scored Critic/Editor improvement loop is internally self-definitional, and it is not load-bearing for the main conclusions.

  1. self definitional [Methods 4.3 (Route criticism and improvement; reported in Section 2.5, Fig. 5b,c)]
    "All three agents use the same backbone as the planner. The loop is an internal consistency procedure, not an external validation: the agent that scores the improvement shares a backbone with the agents that produce it, so a declining blocking rate demonstrates convergence against the Critic’s criterion and not experimental feasibility."

    The improvement-loop result—blocking rate falling from about 0.27 to about 0.06 and the shift in Analyst feasibility scores—is generated by an Editor that repairs routes against the Critic’s flags and then rescored by an Analyst sharing the same LLM backbone. The paper itself reduces the claimed improvement to convergence against the Critic’s own criterion: the same model family supplies both the proposal and the ground truth, so the measured improvement is defined by the Critic’s judgment rather than by any external chemical validation.

full rationale

No load-bearing circularity is present in the paper’s central derivation chains. The 63.9% solve-rate claim is benchmarked against the external open-source planner AiZynthFinder under the same leaf-purchasability criterion and the same ZINC/eMolecules stock; the strategic-layer-only result already exceeds the exhaustive template baseline, so the stitched 63.9% is not forced by construction. The reaction-space distinctness claim is supported by external classifiers—NameRXN is corpus-independent and at parity, and RetroChimera is an external single-step model—while the authors’ own ReactionClassifier is explicitly described as a purpose-built diagnostic whose calibration is disclosed. The expert key-step evaluation is blinded, uses external human raters, and its conditioning on a shared strategic frame is stated as a limitation rather than hidden; that conditioning is a selection criterion, not a reduction of the rating outcome to the model’s own output. The only self-referential element is the Critic/Editor/Analyst improvement loop, which the paper itself concedes is an internal consistency procedure rather than external validation; this is a contained self-definitional loop but it is not used to support the reach or expert-comparison conclusions. Because the central claims retain independent external content, the appropriate score is low, reflecting only this disclosed internal self-scoring loop.

Assumptions & free parameters 6 free parameters · 6 assumptions · 2 invented entities

The benchmark and evaluation rest on external database choices (NP-Atlas, ZINC/eMolecules), on a documented but unverifiable model knowledge cutoff, and on the LLM's own judgment for strategy congruence and step feasibility. These are disclosed by the authors; the central reach comparison is against an external planner, so the core conclusion does not reduce to a fitted parameter.

free parameters (6)
  • LLM sampling temperatures = strategy 0.1; next-step 0.3
    Chosen hyperparameters controlling exploration; not fitted to data, but the reported routes depend on them and generations are non-deterministic (Methods 4.2).
  • Strategies per target = 3
    Default number of strategies generated per target; affects coverage and cost (Methods 4.2).
  • Benchmark complexity filters = Bertz 900-2200; 24-65 heavy atoms; 4-12 stereocenters; 2-8 rings; <=14 rotatable bonds
    Selection thresholds defining 'complex' natural products; changing them changes the benchmark and solve rates (Methods 4.1).
  • Baseline search budgets = depth 25, 1500 iterations, 1800s; leaf completion depth 6, 500 iterations, 1200s
    Compute budgets chosen to make the baseline 'generous' and the completion 'short'; the reach comparison depends on these choices (Methods 4.4).
  • Congruence criterion = LLM judgment (no threshold specified)
    The expert evaluation retains only targets where an LLM judges SynthEx's strategy congruent with the published one (47 of 70); the criterion is not operationalized numerically, so it is a hand-set selection rule (Methods 4.10).
  • Top-k recovery threshold = top-1, top-5, top-50
    Chosen reporting thresholds for RetroChimera recovery (Methods 4.7).
assumptions (6)
  • domain assumption NP-Atlas release 2024-09 correctly flags natural products with no reported total synthesis.
    The benchmark of 1,098 targets is built on this; if NP-Atlas is incomplete, some targets may actually have literature routes, weakening the 'no retrieval' argument (Methods 4.1).
  • domain assumption Gemini 3.1 pro preview has a January 2025 knowledge cutoff and post-training did not include later chemical literature.
    The Okaramine M 'recovery' and the 47 congruent expert-evaluation targets rely on the model not having seen the published routes; the paper itself acknowledges the cutoff is not proof (Methods 4.2).
  • domain assumption Presence of an InChIKey in the ZINC+eMolecules stock (39,684,411 keys) is an adequate proxy for purchasability.
    The solve criterion uses this stock; commercial availability changes over time and the stock may not reflect actual purchasability (Methods 4.4).
  • domain assumption NameRXN naming and expert key-step ratings are valid proxies for chemical soundness of proposed steps.
    The paper uses NameRXN recognition as evidence that SynthEx chemistry is nameable, and expert ratings as the quality signal; neither is experimental validation, and the paper says so (Sections 2.3, 2.4, 3).
  • domain assumption The LLM (Gemini) reliably executes ReactionJSON graph edits and the LLM critic (same backbone) correctly identifies infeasible steps.
    The entire pipeline assumes the model writes correct atom-level edits and that the critic's 'blocking' judgments are meaningful; the improvement loop is self-scored (Methods 4.2, 4.3).
  • standard math RDKit ring-counting and the three-heavy-atom filter correctly classify ring-forming reactions.
    Used to define the ring-forming signature (16.0% vs USPTO); a minor structural assumption (Methods 4.8).
invented entities (2)
  • ReactionJSON independent evidence
    purpose: A template-free reaction representation as ordered atom-level graph edits; the core mechanism letting the LLM propose novel disconnections.
    A data format, not a physical entity. Evidence is the released SynthAtlas corpus and the deterministic application of edits to mapped products; the format itself is defined in Methods 4.2 and the code is open-sourced.
  • RouteJSON independent evidence
    purpose: Linear serialization of a route for surgical editing by the Critic/Editor loop.
    A document format enabling in-place route edits; supported by the Monascuspirolide A case study and released code.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Strategy-first synthesis planning for complex natural products." pith.science (2026). https://pith.science/paper/IAOKL5B7

@misc{pith2026260807454,
  author       = {Pith},
  title        = {Pith review of: Strategy-first synthesis planning for complex natural products},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IAOKL5B7}},
  note         = {Machine review of arXiv:2608.07454}
}
read the original abstract

The total synthesis of a complex molecule is among the most demanding intellectual and experimental feats in chemistry: a chemist must plan many steps ahead for how to assemble simple building blocks into an intricate target, devise backup strategies, and anticipate procedural challenges. It is also a profoundly creative activity. For half a century, efforts to automate the retrosynthetic design of natural products and other complex molecules have drawn on catalogued reactions, and the resulting tools now report near-complete success on benchmarks built from that same source. But these tools were shaped to fit benchmarked chemistry, and they falter on many natural products, the frontier of the field, whose densely functionalized, polycyclic architectures demand precisely the inventive chemistry the record contains least. Whether a machine could reasonably design such syntheses like an expert chemist does has remained unclear. Here, we show that SynthEx, an agentic framework built on large language models, plans routes to complex natural products that lie beyond the reach of conventional design algorithms. SynthEx proposes competing strategies, assembles a sequence of routine and key steps into a cohesive route, and critiques and improves its own design; the chemistry it favours is more convergent than existing tools produce, and spans a region of reaction space that catalogue-based tools cannot match. Most notably, in blinded assessments, expert chemists judged its key steps comparable to those of published human syntheses and engaged with them as genuine synthesis plans, a response algorithmic route prediction has not previously accomplished. We release routes to more than a thousand natural products as SynthAtlas, an open, interactive database, and anticipate it will become a shared resource for a collection of complex target molecules that lack existing literature routes.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

76 extracted references · 62 canonical work pages

  1. [1]

    Journal of the American Chemical Society86(3), 478–485 (1964)

    Corey, E., Ohno, M., Mitra, R.B., Vatakencherry, P.A.: Total synthesis of longifolene. Journal of the American Chemical Society86(3), 478–485 (1964)

    work page 1964

  2. [2]

    John Wiley & Sons, Weinheim (1995)

    Corey, E.J., Cheng, X.-M.: The Logic of Chemical Synthesis. John Wiley & Sons, Weinheim (1995)

    work page 1995

  3. [3]

    Nicolaou, K.C., Sorensen, E.J.: Classics in total synthesis: targets, strategies, methods (1996)

    work page 1996

  4. [4]

    Angewandte Chemie International Edition39, 44–122 (2000)

    Nicolaou, K.C., Vourloumis, D., Winssinger, N., Baran, P.S.: The art and science of total synthesis at the dawn of the twenty-first century. Angewandte Chemie International Edition39, 44–122 (2000)

    work page 2000

  5. [5]

    Trost, B.M., Fleming, I.: Comprehensive Organic Synthesis: Selectivity, Strategy, and Efficiency in Modern Organic Chemistry vol. 8. Elsevier, Oxford (1991)

    work page 1991

  6. [6]

    Liebigs Annalen1997(7), 1283–1301 (1997)

    Nicolaou, K., H¨ arter, M.W., Gunzner, J.L., Nadin, A.: The wittig and related reactions in natural product synthesis. Liebigs Annalen1997(7), 1283–1301 (1997)

    work page 1997

  7. [7]

    Elsevier, Amsterdam (2021)

    Heravi, M.M., Zadsirjan, V.: Recent Applications of Selected Name Reactions in the Total Synthesis of Alkaloids. Elsevier, Amsterdam (2021)

    work page 2021

  8. [8]

    Matter9(5) (2026)

    Bran, A.M., Neukomm, T.A., Armstrong, D., Jonˇ cev, Z., Schwaller, P.: Chem- ical reasoning in llms unlocks strategy-aware synthesis planning and reaction 28 mechanism elucidation. Matter9(5) (2026)

    work page 2026

Show all 76 references
  1. [9]

    Science166(3902), 178–192 (1969)

    Corey, E.J., Wipke, W.T.: Computer-assisted design of complex organic syn- theses: Pathways for molecular synthesis can be devised with a computer and equipment for graphical communication. Science166(3902), 178–192 (1969)

    1969

  2. [10]

    facile man-machine communication of chemical structure by interactive computer graphics

    Corey, E., Wipke, W.T., Cramer III, R.D., Howe, W.J.: Computer-assisted syn- thetic analysis. facile man-machine communication of chemical structure by interactive computer graphics. J. Am. Chem. Soc.94, 421–430 (1972)

    1972

  3. [11]

    Science228, 408–418 (1985)

    Corey, E.J., Long, A.K., Rubenstein, S.D.: Computer-assisted analysis in organic synthesis. Science228, 408–418 (1985)

    1985

  4. [12]

    ACS Publications (2011)

    Lowe, D.M., Corbett, P.T., Murray-Rust, P., Glen, R.C.: Chemical name to structure: OPSIN, an open source solution. ACS Publications (2011)

    2011

  5. [13]

    PhD thesis, University of Cambridge (2012)

    Lowe, D.M.: Extraction of chemical structures and reactions from the literature. PhD thesis, University of Cambridge (2012)

    2012

  6. [14]

    6084/m9.figshare.5104873.v1 (2017)

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

    2017

  7. [15]

    Segler, M.H., Waller, M.P.: Neural-symbolic machine learning for retrosynthesis and reaction prediction. Chem. Eur. J.23, 5966–5971 (2017)

    2017

  8. [16]

    ACS Cent

    Coley, C.W., Rogers, L., Green, W.H., Jensen, K.F.: Computer-assisted retrosyn- thesis based on molecular similarity. ACS Cent. Sci.3, 1237–1245 (2017)

    2017

  9. [17]

    Nature555(7698), 604–610 (2018)

    Segler, M.H., Preuss, M., Waller, M.P.: Planning chemical syntheses with deep neural networks and symbolic ai. Nature555(7698), 604–610 (2018)

    2018

  10. [18]

    Found in Transla- tion

    Schwaller, P., Gaudin, T., Lanyi, D., Bekas, C., Laino, T.: “Found in Transla- tion”: predicting outcomes of complex organic chemistry reactions using neural sequence-to-sequence models. Chem. Sci.9, 6091–6098 (2018)

    2018

  11. [19]

    In: Guyon, I., Luxburg, U.V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., Garnett, R

    Jin, W., Coley, C., Barzilay, R., Jaakkola, T.: Predicting organic reaction out- comes with weisfeiler-lehman network. In: Guyon, I., Luxburg, U.V., Bengio, S., Wallach, H., Fergus, R., Vishwanathan, S., Garnett, R. (eds.) Advances in Neu- ral Information Processing Systems 30...

    2017

  12. [20]

    https://arxiv.org/abs/2006.15820

    Chen, B., Li, C., Dai, H., Song, L.: Retro*: Learning Retrosynthetic Planning with Neural Guided A* Search (2020). https://arxiv.org/abs/2006.15820

    2020 arXiv

  13. [21]

    Sacha, M., Blaz, M., Byrski, P., Dabrowski-Tumanski, P., Chrominski, M., Loska, R., Wlodarczyk-Pruszynski, P., Jastrzebski, S.: Molecule edit graph attention network: Modeling chemical reactions as sequences of graph edits. J. Chem. Inf. Model.61, 3273–3284 (2021) https://doi....

    2021 doi

  14. [22]

    Schwaller, P., Petraglia, R., Zullo, V., Nair, V.H., Haeuselmann, R.A., Pisoni, R., Bekas, C., Iuliano, A., Laino, T.: Predicting retrosynthetic pathways using transformer-based models and a hyper-graph exploration strategy. Chem. Sci.11, 3316–3325 (2020)

    2020

  15. [23]

    In: International Conference on Machine Learning, pp

    Liu, G., Xue, D., Xie, S., Xia, Y., Tripp, A., Maziarz, K., Segler, M., Qin, T., Zhang, Z., Liu, T.-Y.: Retrosynthetic planning with dual value networks. In: International Conference on Machine Learning, pp. 22266–22276 (2023). PMLR

    2023

  16. [24]

    arXiv preprint arXiv:2412.05269 (2024)

    Maziarz, K., Liu, G., Misztela, H., Kornev, A., Gai´ nski, P., Hoefling, H., Fortu- nato, M., Gupta, R., Segler, M.: Chimera: Accurate retrosynthesis prediction by ensembling models with diverse inductive biases. arXiv preprint arXiv:2412.05269 (2024)

    2024 arXiv

  17. [25]

    Journal of Chemical Information and Modeling63(7), 1841–1846 (2023)

    Genheden, S., Norrby, P.-O., Engkvist, O.: Aizynthtrain: robust, reproducible, and extensible pipelines for training synthesis prediction models. Journal of Chemical Information and Modeling63(7), 1841–1846 (2023)

    2023

  18. [26]

    ACS Cent

    Schwaller, P., Laino, T., Gaudin, T., Bolgar, P., Hunter, C.A., Bekas, C., Lee, A.A.: Molecular transformer: a model for uncertainty-calibrated chemical reaction prediction. ACS Cent. Sci.5(9), 1572–1583 (2019)

    2019

  19. [27]

    Digital Discovery5(2), 793–802 (2026)

    Tanovic, S., Wieczorek, E., Duarte, F.: An exploration of dataset bias in single- step retrosynthesis prediction. Digital Discovery5(2), 793–802 (2026)

    2026

  20. [28]

    2024, pp

    Jimenez, C.E., Yang, J., Wettig, A., Yao, S., Pei, K., Press, O., Narasimhan, K.: Swe-bench: Can language models resolve real-world github issues? In: Inter- national Conference on Learning Representations, vol. 2024, pp. 54107–54157 (2024)

    2024

  21. [29]

    Advances in Neural Information Processing Systems37, 50528–50652 (2024)

    Yang, J., Jimenez, C., Wettig, A., Lieret, K., Yao, S., Narasimhan, K., Press, O.: Swe-agent: Agent-computer interfaces enable automated software engineering. Advances in Neural Information Processing Systems37, 50528–50652 (2024)

    2024

  22. [30]

    ChemRxiv (2022)

    Genheden, S., Bjerrum, E.: Paroutes: a framework for benchmarking retrosyn- thesis route predictions. ChemRxiv (2022)

    2022

  23. [31]

    arXiv preprint arXiv:2310.19796 (2023)

    Maziarz, K., Tripp, A., Liu, G., Stanley, M., Xie, S., Gai´ nski, P., Seidl, P., Segler, M.: Re-evaluating retrosynthesis algorithms with syntheseus. arXiv preprint arXiv:2310.19796 (2023)

    2023 arXiv

  24. [32]

    In: International Confer- ence on Learning Representations, vol

    Tripp, A., Maziarz, K., Lewis, S., Segler, M., Hern´ andez Lobato, J.M.: Retro- fallback: retrosynthetic planning in an uncertain world. In: International Confer- ence on Learning Representations, vol. 2024, pp. 30687–30744 (2024) 30

    2024

  25. [33]

    arXiv preprint arXiv:2512.07079 (2025)

    Morgunov, A., Batista, V.S.: Procrustean bed for ai-driven retrosynthesis: A unified framework for reproducible evaluation. arXiv preprint arXiv:2512.07079 (2025)

    2025

  26. [34]

    arXiv preprint arXiv:2507.21762 (2025)

    Xuan-Vu, N., Armstrong, D., Joncev, Z., Schwaller, P.: Tempre: Template generation for single and direct multi-step retrosynthesis. arXiv preprint arXiv:2507.21762 (2025)

    2025 arXiv

  27. [35]

    Digital Discovery5(4), 1623–1634 (2026) https://doi.org/10.1039/d5dd00280j https://pubs.rsc.org/dd/article-pdf/5/4/1623/12756083/d5dd00280j.pdf

    Hassen, A.K., Lai, H., Genheden, S., Preuss, M., Clevert, D.-A.: Synthe- sis planning in reaction space: a study on success, robustness and diversity. Digital Discovery5(4), 1623–1634 (2026) https://doi.org/10.1039/d5dd00280j https://pubs.rsc.org/dd/article-pdf/5/4/1623/127560...

    2026 doi

  28. [36]

    Chemical Science (2026)

    Tran, S.B., Roh, J., Coley, C.W.: Quantifying the failure modes of current one- step retrosynthesis models. Chemical Science (2026)

    2026

  29. [37]

    ChemRxiv (2026) https://doi.org/10.26434/chemrxiv.15001278

    Morgunov, A., Shee, Y., Soudackov, A.V., Batista, V.S.: The syntax of matter: Synthesis planning as the foundation of generative chemistry. ChemRxiv (2026) https://doi.org/10.26434/chemrxiv.15001278

    2026 doi

  30. [38]

    Digital discovery2(5), 1233–1250 (2023)

    Jablonka, K.M., Ai, Q., Al-Feghali, A., Badhwar, S., Bocarsly, J.D., Bran, A.M., Bringuier, S., Brinson, L.C., Choudhary, K., Circi, D.,et al.: 14 examples of how llms can transform materials science and chemistry: a reflection on a large language model hackathon. Digital disc...

    2023

  31. [39]

    Jablonka, K.M., Schwaller, P., Ortega-Guerrero, A., Smit, B.: Leveraging large language models for predictive chemistry. Nat. Mach. Intell.6(2), 161–169 (2024) https://doi.org/10.1038/s42256-023-00788-1

    2024 doi

  32. [40]

    arXiv preprint arXiv:2411.16955 (2024)

    Alampara, N., Schilling-Wilhelmi, M., R´ ıos-Garc´ ıa, M., Mandal, I., Khetarpal, P., Grover, H.S., Krishnan, N., Jablonka, K.M.: Probing the limitations of mul- timodal language models for chemistry and materials research. arXiv preprint arXiv:2411.16955 (2024)

    2024 arXiv

  33. [41]

    Nature Machine Intelligence, 1–11 (2024)

    Bran, A., Cox, S., Schilter, O., Baldassari, C., White, A.D., Schwaller, P.: Aug- menting large language models with chemistry tools. Nature Machine Intelligence, 1–11 (2024)

    2024

  34. [42]

    Nature624(7992), 570–578 (2023)

    Boiko, D.A., MacKnight, R., Kline, B., Gomes, G.: Autonomous chemical research with large language models. Nature624(7992), 570–578 (2023)

    2023

  35. [43]

    Nature Chemistry, 1–8 (2025)

    Mirza, A., Alampara, N., Kunchapu, S., R´ ıos-Garc´ ıa, M., Emoekabu, B., Krish- nan, A., Gupta, T., Schilling-Wilhelmi, M., Okereke, M., Aneesh, A., et al.: A framework for evaluating the chemical knowledge and reasoning abilities of large language models against the expertis...

    2025

  36. [44]

    arXiv preprint arXiv:2512.01507 (2025)

    Armstrong, D., Jonˇ cev, Z., Bran, A.M., Schwaller, P.: Synthstrategy: Extracting 31 and formalizing latent strategic insights from llms in organic chemistry. arXiv preprint arXiv:2512.01507 (2025)

    2025

  37. [45]

    arXiv preprint arXiv:2508.11860 (2025)

    Baker, F.N., Adu-Ampratwum, D., Averly, R., Yu, B., Sun, H., Ning, X.: Larc: Towards human-level constrained retrosynthesis planning through an agentic framework. arXiv preprint arXiv:2508.11860 (2025)

    2025 arXiv

  38. [46]

    arXiv preprint arXiv:2604.05075 (2026)

    Baker, F.N., Nguyen, T., Averly, R., Yu, B., Adu-Ampratwum, D., Sun, H., Ning, X.: Mmorf: A multi-agent framework for designing multi-objective retrosynthesis planning systems. arXiv preprint arXiv:2604.05075 (2026)

    2026 arXiv

  39. [47]

    arXiv preprint arXiv:2512.16424 (2025)

    Xuan-Vu, N., Armstrong, D., Wehrbach, M., Bran, A.M., Jonˇ cev, Z., Schwaller, P.: Synthelite: Chemist-aligned and feasibility-aware synthesis planning with llms. arXiv preprint arXiv:2512.16424 (2025)

    2025

  40. [48]

    (Accessed Jul 29, 2021) (2024)

    Reaxys database. (Accessed Jul 29, 2021) (2024). https://www.reaxys.com

    2024

  41. [49]

    Nature596(7873), 583–589 (2021)

    Jumper, J., Evans, R., Pritzel, A., Green, T., Figurnov, M., Ronneberger, O., Tunyasuvunakool, K., Bates, R., ˇZ´ ıdek, A., Potapenko, A.,et al.: Highly accurate protein structure prediction with alphafold. Nature596(7873), 583–589 (2021)

    2021

  42. [50]

    arXiv preprint arXiv:2605.01822 (2026)

    Runcie, N.T., Imrie, F., Deane, C.M.: Molecular representations for large language models. arXiv preprint arXiv:2605.01822 (2026)

    2026 arXiv

  43. [51]

    Organic Letters27(27), 7367–7371 (2025)

    Schmelzer, D., Stark, C.B.: Synthesis of Okaramine M, Its Conversion to Amauromines, and Concise Bidirectional and Biomimetic Total Synthesis of Amauromines. Organic Letters27(27), 7367–7371 (2025)

    2025

  44. [52]

    Tetrahedron Letters25(41), 4673–4676 (1984)

    Takase, S., Kawai, Y., Uchida, I., Tanaka, H., Aoki, H.: Structure of amau- romine, a new alkaloid with vasodilating activity produced by amauroascus sp. Tetrahedron Letters25(41), 4673–4676 (1984)

    1984

  45. [53]

    Chemical and Pharmaceutical Bulletin58(5), 717–719 (2010)

    Ishikawa, K., Hosoe, T., Itabashi, T., Wakana, D., Takizawa, K., Yaguchi, T., Kawai, K.-i.: Novoamauromine and ent-Cycloechinulin: two new diketopiper- azine derivatives from Aspergillus novofumigatus. Chemical and Pharmaceutical Bulletin58(5), 717–719 (2010)

    2010

  46. [54]

    Organic Letters23(15), 5964–5968 (2021)

    Kouam´ e, T., Bernadat, G., Turpin, V., Litaudon, M., Okpekon, A.T., Gallard, J.-F., Leblanc, K., Rharrabti, S., Champy, P., Poupon, E.,et al.: Structure Reas- signment of Melonine and Quantum-Chemical Calculations-Based Assessment of Biosynthetic Scenarios Leading to Its Revi...

    2021

  47. [55]

    Natural Product Reports41(5), 784– 812 (2024) 32

    Zhang, X.: Vallesamidine and schizozygane alkaloids: rearranged monoterpene indole alkaloids and synthetic endeavours. Natural Product Reports41(5), 784– 812 (2024) 32

    2024

  48. [56]

    Organic Letters27(9), 2065–2068 (2025)

    Matsuyuki, Y., Umekubo, N., Yokoshima, S.: Total Synthesis of Melonine. Organic Letters27(9), 2065–2068 (2025)

    2025

  49. [57]

    Angewandte Chemie International Edition, 8101956 (2026)

    Go¨ elo, V., Wang, Q., Zhu, J.: Total Synthesis of (+)-Melonine and (+)-N 4- Oxy Melonine Enabled by an Intramolecular Alkene Diamination Reaction. Angewandte Chemie International Edition, 8101956 (2026)

    2026

  50. [58]

    Angewandte Chemie International Edition59(33), 13990–13997 (2020) https://doi.org/10.1002/anie.202005380 https://onlinelibrary.wiley.com/doi/pdf/10.1002/anie.202005380

    Delayre, B., Piemontesi, C., Wang, Q., Zhu, J.: TiCl3-Mediated Synthesis of 2,3,3- Trisubstituted Indolenines: Total Synthesis of (+)-1,2-Dehydroaspidospermidine, (+)-Condyfoline, and (−)-Tubifoline. Angewandte Chemie International Edition59(33), 13990–13997 (2020) https://doi...

    2020 doi

  51. [59]

    Frontiers in Bioengineering and Biotechnology10, 1095464 (2022)

    Ma, Y., Yan, J., Yang, L., Yao, Y., Wang, L., Gao, S.-S., Cui, C.: A hybrid system for the overproduction of complex ergot alkaloid chanoclavine. Frontiers in Bioengineering and Biotechnology10, 1095464 (2022)

    2022

  52. [60]

    Tetrahedron Letters14(29), 2699–2701 (1973)

    Dupeyre, R.-M., Rassat, A.: Application de la reaction de Hofmann-L¨ offler- Freytag synthese de derives diaza-2, 6 adamantane. Tetrahedron Letters14(29), 2699–2701 (1973)

    1973

  53. [61]

    Natural Product Reports31(10), 1328–1338 (2014)

    Jakubczyk, D., Cheng, J.Z., O’Connor, S.E.: Biosynthesis of the ergot alkaloids. Natural Product Reports31(10), 1328–1338 (2014)

    2014

  54. [62]

    Nextmove Software NameRXN http://www.nextmovesoftware.com/namerxn. html. (Accessed March 29, 2022). http://www.nextmovesoftware.com/namerxn. html

    2022

  55. [63]

    Journal of Cheminfor- matics16(1), 37 (2024)

    Dobbelaere, M.R., Lengyel, I., Stevens, C.V., Van Geem, K.M.: Rxn-insight: fast chemical reaction analysis using bond-electron matrices. Journal of Cheminfor- matics16(1), 37 (2024)

    2024

  56. [64]

    arXiv preprint arXiv:2607.01061 (2026)

    Armstrong, D., Dobbelaere, M., Olikauskas, V., Avila, H., Susanu, O., Waser, J., Schwaller, P.: Agentic generation of verifiable rules for deterministic, self- expanding reaction classification. arXiv preprint arXiv:2607.01061 (2026)

    2026 arXiv

  57. [65]

    arXiv preprint arXiv:2407.06334 (2024)

    Yu, K., Roh, J., Li, Z., Gao, W., Wang, R., Coley, C.W.: Double-ended synthesis planning with goal-constrained bidirectional search. arXiv preprint arXiv:2407.06334 (2024)

    2024 arXiv

  58. [66]

    Genheden, S., Thakkar, A., Chadimov´ a, V., Reymond, J.-L., Engkvist, O., Bjer- rum, E.: AiZynthFinder: a fast, robust and flexible open-source software for retrosynthetic planning. J. Cheminf.12, 1–9 (2020)

    2020

  59. [67]

    Advances in neural information processing systems36, 8634–8652 (2023) 33

    Shinn, N., Cassano, F., Gopinath, A., Narasimhan, K., Yao, S.: Reflexion: Lan- guage agents with verbal reinforcement learning. Advances in neural information processing systems36, 8634–8652 (2023) 33

    2023

  60. [68]

    In: 2025 IEEE/ACM 47th International Conference on Software Engineering: Software Engineering in Practice (ICSE-SEIP), pp

    Takerngsaksiri, W., Pasuksmit, J., Thongtanunam, P., Tantithamthavorn, C., Zhang, R., Jiang, F., Li, J., Cook, E., Chen, K., Wu, M.: Human-in-the-loop software development agents. In: 2025 IEEE/ACM 47th International Conference on Software Engineering: Software Engineering in ...

    2025

  61. [69]

    arXiv preprint arXiv:2505.07027 (2025)

    Wang, H., Guo, J., Kong, L., Ramprasad, R., Schwaller, P., Du, Y., Zhang, C.: Llm-augmented chemical synthesis and design decision programs. arXiv preprint arXiv:2505.07027 (2025)

    2025 arXiv

  62. [70]

    Edwards, C., Lai, T., Ros, K., Honke, G., Cho, K., Ji, H.: Translation between molecules and natural language. Proc. Conf. Empirical Methods Nat. Lang. Process., 375–413 (2022)

    2022

  63. [71]

    http://practicalcheminformatics.blogspot.com/2024/10/ silly-things-large-language-models-do.html Accessed 2025-02-03

    Walters, P.: Silly Things Large Language Models Do With Molecules. http://practicalcheminformatics.blogspot.com/2024/10/ silly-things-large-language-models-do.html Accessed 2025-02-03

    2024

  64. [72]

    Jang, H., Jang, Y., Kim, J., Ahn, S.: Can LLMs Generate Diverse Molecules? Towards Alignment with Structural Diversity (2024)

    2024

  65. [73]

    Journal of Chemical Information and Modeling66(11), 6361–6375 (2026)

    Westerlund, A.M., Sigmund, L.M., Mijangos, M.V., Kannas, C., Genheden, S., Kabeshov, M.: Toward lab-ready ai synthesis plans with protection strategies and route scoring. Journal of Chemical Information and Modeling66(11), 6361–6375 (2026)

    2026

  66. [74]

    Digital Discovery4(4), 1042–1051 (2025)

    P´ erez, K.L., Jung, V., Chen, L., Huddleston, K., Miranda-Quintana, R.A.: Bit- birch: efficient clustering of large molecular libraries. Digital Discovery4(4), 1042–1051 (2025)

    2025

  67. [75]

    Gemini API documentation

    Google: Gemini 3 models. Gemini API documentation. https://ai.google.dev/ gemini-api/docs/gemini-3. Knowledge cutoff January 2025. Accessed 25 July 2026 (2026)

    2026

  68. [76]

    would I run this

    Schwaller, P., Probst, D., Vaucher, A.C., Nair, V.H., Kreutter, D., Laino, T., Reymond, J.-L.: Mapping the space of chemical reactions using attention-based neural networks. Nat. Mach. Intell.3, 144–152 (2021) A Assembling the blind chemist evaluation set We asked expert chemi...

    2021

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.