Quantum-inspired Reinforcement Learning for Synthesizable Drug Design

Dannong Wang; Jintai Chen; Lulu Chen; Minjie Shen; Tianfan Fu; Xiao-Yang Liu; Yingzhou Lu; Zhiding Liang

arxiv: 2409.09183 · v2 · submitted 2024-09-13 · 💻 cs.LG · q-bio.BM

Quantum-inspired Reinforcement Learning for Synthesizable Drug Design

Dannong Wang , Jintai Chen , Yingzhou Lu , Minjie Shen , Lulu Chen , Zhiding Liang , Tianfan Fu , Xiao-Yang Liu This is my paper

Pith reviewed 2026-05-23 20:21 UTC · model grok-4.3

classification 💻 cs.LG q-bio.BM

keywords reinforcement learningmolecular optimizationdrug designsimulated annealinggenetic algorithmssynthesizable moleculespolicy network

0 comments

The pith

Reinforcement learning with a quantum-inspired simulated annealing policy network guides transitions in chemical space for synthesizable molecular design.

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

The paper presents a method that trains a policy neural network with deterministic REINFORCE to output probabilities for moving between molecular states, then applies genetic algorithm local search to refine each candidate to a local optimum. This combination is intended to navigate the enormous discrete space of chemical structures more effectively than random search while respecting synthetic feasibility constraints. The approach is tested inside the Practical Molecular Optimization benchmark using a strict 10K query budget and is shown to reach performance levels comparable to leading genetic-algorithm baselines. A reader would care because better-directed search could reduce the number of expensive property evaluations needed to discover viable drug candidates.

Core claim

The central claim is that a policy neural network trained by deterministic REINFORCE inside a quantum-inspired simulated annealing schedule can produce useful transitional probabilities that, when paired with genetic-algorithm local search inside each iteration, enable competitive optimization of molecular properties on the PMO benchmark under a 10K-query limit.

What carries the argument

The quantum-inspired simulated annealing policy neural network that outputs transitional probabilities to guide state transitions between molecular structures.

If this is right

The method reaches performance comparable to state-of-the-art genetic-algorithm approaches on the PMO benchmark within a 10K-query budget.
Each iteration combines global guidance from the policy network with local refinement by genetic operators to reach local optima.
The approach is designed to scale to the vast discrete space of synthesizable chemical structures rather than relying on exhaustive enumeration.
Deterministic REINFORCE supplies the training signal that updates the network's output probabilities across iterations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

If the performance advantage disappears when the policy network is removed, the reinforcement-learning component would be shown to be the primary driver rather than the genetic-algorithm subroutine.
The same policy-guided transition mechanism could be tested on other discrete combinatorial optimization tasks that share the structure of large state spaces and expensive evaluation oracles.
Extending the query budget or replacing the oracle functions with more realistic multi-objective drug-discovery criteria would reveal whether the observed competitiveness holds under different resource constraints.

Load-bearing premise

The transitional probabilities produced by the trained policy network actually improve the search beyond what the embedded genetic-algorithm local search and the benchmark oracles would achieve on their own.

What would settle it

An ablation that replaces the learned policy network with uniform random transition probabilities and measures whether the resulting performance on the PMO benchmark falls to or below that of the pure genetic-algorithm baseline.

read the original abstract

Synthesizable molecular design (also known as synthesizable molecular optimization) is a fundamental problem in drug discovery, and involves designing novel molecular structures to improve their properties according to drug-relevant oracle functions (i.e., objective) while ensuring synthetic feasibility. However, existing methods are mostly based on random search. To address this issue, in this paper, we introduce a novel approach using the reinforcement learning method with quantum-inspired simulated annealing policy neural network to navigate the vast discrete space of chemical structures intelligently. Specifically, we employ a deterministic REINFORCE algorithm using policy neural networks to output transitional probability to guide state transitions and local search using genetic algorithm to refine solutions to a local optimum within each iteration. Our methods are evaluated with the Practical Molecular Optimization (PMO) benchmark framework with a 10K query budget. We further showcase the competitive performance of our method by comparing it against the state-of-the-art genetic algorithms-based method.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

Hybrid RL-GA method for molecular design lacks ablations to show if the policy adds value over the genetic search.

read the letter

Colleague, the core of this paper is a hybrid approach to synthesizable molecular optimization that trains a policy network with deterministic REINFORCE, draws on quantum-inspired simulated annealing for the policy, and runs genetic algorithm local search inside each iteration to refine candidates. They evaluate on the PMO benchmark with a 10K query budget and position the results as competitive with existing GA-based methods. That combination is the main thing to note. The work does apply RL ideas to guide transitions in chemical space rather than relying solely on random or evolutionary search, and the choice of the PMO framework is appropriate for the domain. The abstract frames the quantum-inspired policy as the element that navigates intelligently, which is a reasonable direction to explore. The main limitation is that no performance numbers, error bars, or ablation results appear in the abstract, so the claim of competitiveness cannot be checked directly. The stress-test concern holds: without experiments that isolate the policy network's contribution from the GA refinement step, it remains possible that the local search or the benchmark oracles account for most of the reported behavior. The full paper may contain those controls and the actual metrics, but on the information given the evidence for the RL component's role is thin. This is aimed at people working on RL or evolutionary methods for molecular design. A reader already following PMO-style benchmarks could pick up the hybrid idea, but the lack of detailed results limits how much can be taken from it right now. It deserves peer review because the domain matters and the setup is a plausible extension of existing work; referees can ask for the missing ablations and numbers. Recommendation: send to review rather than desk reject, with the expectation that the authors will need to demonstrate the policy's incremental value.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a reinforcement learning method for synthesizable drug design that employs a quantum-inspired simulated annealing policy neural network with deterministic REINFORCE to output transitional probabilities for state transitions, augmented by genetic algorithm local search within each iteration. The method is tested on the Practical Molecular Optimization (PMO) benchmark using a 10K query budget and is asserted to achieve competitive performance relative to state-of-the-art genetic algorithm approaches.

Significance. Should the empirical claims be confirmed with rigorous ablations demonstrating the policy's contribution, this approach could meaningfully advance the application of RL techniques in navigating discrete molecular spaces for drug discovery, offering potential improvements over purely GA-based methods.

major comments (2)

[Methods] Methods: The integration of the policy neural network with GA local search is described, but no ablation experiments are mentioned to isolate the contribution of the learned transitional probabilities from the GA refinement. This is load-bearing for the central claim that the quantum-inspired RL 'navigates the vast discrete space of chemical structures intelligently' rather than the performance being driven primarily by the embedded GA.
[Experiments] Experiments: The evaluation on PMO with 10K budget claims competitive performance, but the abstract provides no quantitative results, error bars, or specific comparison metrics, preventing verification of the claim against SOTA GA methods.

minor comments (2)

[Abstract] Abstract: The phrase 'quantum-inspired simulated annealing policy neural network' is used without defining how the quantum inspiration is realized in the policy network architecture or training.
[Abstract] Abstract: Clarify the specific components of the PMO benchmark used, including the oracle functions, to allow reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments. We address each major point below and indicate the revisions we will make to strengthen the manuscript.

read point-by-point responses

Referee: [Methods] Methods: The integration of the policy neural network with GA local search is described, but no ablation experiments are mentioned to isolate the contribution of the learned transitional probabilities from the GA refinement. This is load-bearing for the central claim that the quantum-inspired RL 'navigates the vast discrete space of chemical structures intelligently' rather than the performance being driven primarily by the embedded GA.

Authors: We agree that ablation experiments are required to substantiate the contribution of the policy network. In the revised version we will add a dedicated ablation study comparing the full method against a baseline that replaces the learned transitional probabilities with uniform random transitions while retaining the GA local search. This will directly quantify the policy's role in guiding navigation. revision: yes
Referee: [Experiments] Experiments: The evaluation on PMO with 10K budget claims competitive performance, but the abstract provides no quantitative results, error bars, or specific comparison metrics, preventing verification of the claim against SOTA GA methods.

Authors: We acknowledge the abstract lacks numerical detail. We will revise the abstract to report the key quantitative metrics (e.g., mean performance and standard deviation across runs) and explicit comparisons against the GA baselines on the PMO tasks under the 10K budget. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper presents an algorithmic combination of deterministic REINFORCE policy networks and embedded genetic-algorithm local search, evaluated empirically on the PMO benchmark. No equations, fitted parameters, or derivation steps are described that reduce any claimed result to its own inputs by construction. No self-citations, uniqueness theorems, or ansatzes are invoked in a load-bearing manner. The performance claims rest on external benchmark comparisons rather than self-referential reductions, making the derivation chain self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review supplies no explicit free parameters, axioms, or invented entities; the method description invokes standard RL and GA components whose details are not given.

pith-pipeline@v0.9.0 · 5708 in / 1184 out tokens · 32028 ms · 2026-05-23T20:21:24.714699+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

we employ a deterministic REINFORCE algorithm using policy neural networks to output transitional probability to guide state transitions and local search using genetic algorithm
IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

quantum-inspired simulated annealing policy neural network

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

43 extracted references · 43 canonical work pages · 3 internal anchors

[1]

, " * write output.state after.block = add.period write newline

ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

work page
[2]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

work page
[3]

Ahn, S.; Kim, J.; Lee, H.; and Shin, J. 2020. Guiding deep molecular optimization with genetic exploration. Advances in neural information processing systems, 33: 12008--12021

work page 2020
[4]

J.; Lahlou, S.; Tiwari, M.; and Bengio, E

Bengio, Y.; Deleu, T.; Hu, E. J.; Lahlou, S.; Tiwari, M.; and Bengio, E. 2021. GFlowNet Foundations. CoRR, abs/2111.09266

work page arXiv 2021
[5]

Bickerton, R.; Paolini, G.; Besnard, J.; Muresan, S.; and Hopkins, A. 2012. Quantifying the chemical beauty of drugs. Nature chemistry, 4: 90--8

work page 2012
[6]

S.; McMartin, C.; and Guida, W

Bohacek, R. S.; McMartin, C.; and Guida, W. C. 1996. The art and practice of structure-based drug design: a molecular modeling perspective. Medicinal research reviews, 16(1): 3--50

work page 1996
[7]

De Cao and T

Cao, N. D.; and Kipf, T. 2018. MolGAN: An implicit generative model for small molecular graphs. arXiv:1805.11973

work page arXiv 2018
[8]

P.; Yu, G.; Herrington, D

Chang, Y.-T.; Hoffman, E. P.; Yu, G.; Herrington, D. M.; Clarke, R.; Wu, C.-T.; Chen, L.; and Wang, Y. 2019. Integrated identification of disease specific pathways using multi-omics data. bioRxiv, 666065

work page 2019
[9]

Chen, J.; Hu, Y.; Wang, Y.; Lu, Y.; Cao, X.; Lin, M.; Xu, H.; Wu, J.; Xiao, C.; Sun, J.; et al. 2024 a . Trialbench: Multi-modal artificial intelligence-ready clinical trial datasets. arXiv preprint arXiv:2407.00631

work page arXiv 2024
[10]

E.; Herrington, D

Chen, L.; Lu, Y.; Wu, C.-T.; Clarke, R.; Yu, G.; Van Eyk, J. E.; Herrington, D. M.; and Wang, Y. 2021. Data-driven detection of subtype-specific differentially expressed genes. Scientific reports, 11(1): 332

work page 2021
[11]

Chen, T.; Hao, N.; Lu, Y.; and Van Rechem, C. 2024 b . Uncertainty Quantification on Clinical Trial Outcome Prediction. arXiv preprint arXiv:2401.03482

work page arXiv 2024
[12]

V.; Chen, J.; and Fu, T

Chen, T.; Lu, Y.; Hao, N.; Rechem, C. V.; Chen, J.; and Fu, T. 2024 c . Uncertainty quantification and interpretability for clinical trial approval prediction. Health Data Science

work page 2024
[13]

Delahaye, D.; Chaimatanan, S.; and Mongeau, M. 2019. Simulated annealing: From basics to applications. Handbook of metaheuristics, 1--35

work page 2019
[14]

W.; and Sun, J

Fu, T.; Gao, W.; Coley, C. W.; and Sun, J. 2022 a . Reinforced Genetic Algorithm for Structure-based Drug Design. In Annual Conference on Neural Information Processing Systems (NeurIPS)

work page 2022
[15]

W.; and Sun, J

Fu, T.; Gao, W.; Xiao, C.; Yasonik, J.; Coley, C. W.; and Sun, J. 2022 b . Differentiable Scaffolding Tree for Molecular Optimization. International Conference on Learning Representations

work page 2022
[16]

M.; and Sun, J

Fu, T.; Xiao, C.; Li, X.; Glass, L. M.; and Sun, J. 2021. MIMOSA : Multi-constraint Molecule Sampling for Molecule Optimization. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, 125--133

work page 2021
[17]

Gao, W.; Fu, T.; Sun, J.; and Coley, C. W. 2022. Sample Efficiency matters: benchmarking molecular optimization. Neural Information Processing Systems (NeurIPS) Track on Datasets and Benchmarks

work page 2022
[18]

Gao, W.; Mercado, R.; and Coley, C. W. 2022 a . Amortized Tree Generation for Bottom-up Synthesis Planning and Synthesizable Molecular Design. arXiv:2110.06389

work page arXiv 2022
[19]

Gao, W.; Mercado, R.; and Coley, C. W. 2022 b . Amortized Tree Generation for Bottom-up Synthesis Planning and Synthesizable Molecular Design. International Conference on Learning Representations

work page 2022
[20]

M.; Fu, T.; Xiao, C.; and Sun, J

Glass, L. M.; Fu, T.; Xiao, C.; and Sun, J. 2021. MOLER: Incorporate molecule-level reward to enhance deep generative model for molecule optimization. IEEE transactions on knowledge and data engineering, 34(11): 5459--5471

work page 2021
[21]

N.; Duvenaud, D.; Hern \'a ndez-Lobato, J

G \'o mez-Bombarelli, R.; Wei, J. N.; Duvenaud, D.; Hern \'a ndez-Lobato, J. M.; S \'a nchez-Lengeling, B.; Sheberla, D.; Aguilera-Iparraguirre, J.; Hirzel, T. D.; Adams, R. P.; and Aspuru-Guzik, A. 2018. Automatic chemical design using a data-driven continuous representation of molecules. ACS central science, 4(2): 268--276

work page 2018
[22]

Objective-Reinforced Generative Adversarial Networks (ORGAN) for Sequence Generation Models

Guimaraes, G. L.; Sanchez-Lengeling, B.; Outeiral, C.; Farias, P. L. C.; and Aspuru-Guzik, A. 2017. Objective-reinforced generative adversarial networks (ORGAN) for sequence generation models. arXiv preprint arXiv:1705.10843

work page internal anchor Pith review Pith/arXiv arXiv 2017
[23]

Huang, T

Huang, K.; Fu, T.; Gao, W.; Zhao, Y.; Roohani, Y.; Leskovec, J.; Coley, C. W.; Xiao, C.; Sun, J.; and Zitnik, M. 2021. Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Development. arXiv:2102.09548

work page arXiv 2021
[24]

Jensen, J. H. 2019. A graph-based genetic algorithm and generative model/Monte Carlo tree search for the exploration of chemical space. Chemical science, 10(12): 3567--3572

work page 2019
[25]

Jin, W.; Barzilay, R.; and Jaakkola, T. 2018. Junction tree variational autoencoder for molecular graph generation. ICML

work page 2018
[26]

Jin, W.; Barzilay, R.; and Jaakkola, T. 2020. Multi-objective molecule generation using interpretable substructures. In International Conference on Machine Learning, 4849--4859. PMLR

work page 2020
[27]

Korovina, K.; Xu, S.; Kandasamy, K.; Neiswanger, W.; Poczos, B.; Schneider, J.; and Xing, E. 2020. ChemBO : Bayesian optimization of small organic molecules with synthesizable recommendations. In International Conference on Artificial Intelligence and Statistics, 3393--3403. PMLR

work page 2020
[28]

Li, Y.; Zhang, L.; and Liu, Z. 2018. Multi-Objective De Novo Drug Design with Conditional Graph Generative Model. arXiv:1801.07299

work page internal anchor Pith review Pith/arXiv arXiv 2018
[29]

Liu, M.; Yan, K.; Oztekin, B.; and Ji, S. 2021. GraphEBM: Molecular graph generation with energy-based models. arXiv preprint arXiv:2102.00546

work page arXiv 2021
[30]

Lu, Y.; and Liu, X.-Y. 2023. Reinforcement Learning for Ising Model. In Thirty-seventh Conference on Neural Information Processing Systems Track on Machine Learning for Physical Sciences

work page 2023
[31]

J.; Cheng, Z.; Saylor, G.; Van Eyk, J

Lu, Y.; Wu, C.-T.; Parker, S. J.; Cheng, Z.; Saylor, G.; Van Eyk, J. E.; Yu, G.; Clarke, R.; Herrington, D. M.; and Wang, Y. 2022. COT : an efficient and accurate method for detecting marker genes among many subtypes. Bioinformatics Advances, 2(1): vbac037

work page 2022
[32]

Luo, Y.; Yan, K.; and Ji, S. 2021. GraphDF : A discrete flow model for molecular graph generation. Proceedings of the 38th International Conference on Machine Learning, ICML , 139: 7192--7203

work page 2021
[33]

Nigam, A.; Friederich, P.; Krenn, M.; and Aspuru-Guzik, A. 2020. Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space. In The International Conference on Learning Representations (ICLR)

work page 2020
[34]

Olivecrona, M.; Blaschke, T.; Engkvist, O.; and Chen, H. 2017 a . Molecular de-novo design through deep reinforcement learning. Journal of Cheminformatics

work page 2017
[35]

Olivecrona, M.; Blaschke, T.; Engkvist, O.; and Chen, H. 2017 b . Molecular De Novo Design through Deep Reinforcement Learning. CoRR, abs/1704.07555

work page internal anchor Pith review Pith/arXiv arXiv 2017
[36]

Rajak, A.; Suzuki, S.; Dutta, A.; and Chakrabarti, B. K. 2023. Quantum annealing: An overview. Philosophical Transactions of the Royal Society A, 381(2241): 20210417

work page 2023
[37]

Shen, C.; Krenn, M.; Eppel, S.; and Aspuru-Guzik, A. 2021. Deep Molecular Dreaming: Inverse machine learning for de-novo molecular design and interpretability with surjective representations. Machine Learning: Science and Technology

work page 2021
[38]

Shi, C.; Xu, M.; Zhu, Z.; Zhang, W.; Zhang, M.; and Tang, J. 2020. GraphAF : a Flow-based Autoregressive Model for Molecular Graph Generation. In The International Conference on Learning Representations (ICLR)

work page 2020
[39]

Sterling, T.; and Irwin, J. J. 2015. ZINC 15--Ligand Discovery for Everyone. Journal of Chemical Information and Modeling, 55(11): 2324--2337

work page 2015
[40]

Sun, J.; and Fu, T. 2022. Antibody complementarity determining regions (cdrs) design using constrained energy model. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 389--399

work page 2022
[41]

J.; Lu, Y.; Van Eyk, J

Wu, C.-T.; Shen, M.; Du, D.; Cheng, Z.; Parker, S. J.; Lu, Y.; Van Eyk, J. E.; Yu, G.; Clarke, R.; Herrington, D. M.; et al. 2022. Cosbin: cosine score-based iterative normalization of biologically diverse samples. Bioinformatics Advances, 2(1): vbac076

work page 2022
[42]

You, J.; et al. 2018. Graph Convolutional Policy Network for Goal-directed Molecular Graph Generation. In Proceedings of the 32Nd International Conference on Neural Information Processing Systems, 6412--6422. Curran Associates Inc

work page 2018
[43]

N.; and Riley, P

Zhou, Z.; Kearnes, S.; Li, L.; Zare, R. N.; and Riley, P. 2019. Optimization of molecules via deep reinforcement learning. Scientific reports

work page 2019

[1] [1]

, " * write output.state after.block = add.period write newline

ENTRY address archivePrefix author booktitle chapter edition editor eid eprint howpublished institution isbn journal key month note number organization pages publisher school series title type volume year label extra.label sort.label short.list INTEGERS output.state before.all mid.sentence after.sentence after.block FUNCTION init.state.consts #0 'before.a...

work page

[2] [2]

write newline

" write newline "" before.all 'output.state := FUNCTION n.dashify 't := "" t empty not t #1 #1 substring "-" = t #1 #2 substring "--" = not "--" * t #2 global.max substring 't := t #1 #1 substring "-" = "-" * t #2 global.max substring 't := while if t #1 #1 substring * t #2 global.max substring 't := if while FUNCTION word.in bbl.in capitalize " " * FUNCT...

work page

[3] [3]

Ahn, S.; Kim, J.; Lee, H.; and Shin, J. 2020. Guiding deep molecular optimization with genetic exploration. Advances in neural information processing systems, 33: 12008--12021

work page 2020

[4] [4]

J.; Lahlou, S.; Tiwari, M.; and Bengio, E

Bengio, Y.; Deleu, T.; Hu, E. J.; Lahlou, S.; Tiwari, M.; and Bengio, E. 2021. GFlowNet Foundations. CoRR, abs/2111.09266

work page arXiv 2021

[5] [5]

Bickerton, R.; Paolini, G.; Besnard, J.; Muresan, S.; and Hopkins, A. 2012. Quantifying the chemical beauty of drugs. Nature chemistry, 4: 90--8

work page 2012

[6] [6]

S.; McMartin, C.; and Guida, W

Bohacek, R. S.; McMartin, C.; and Guida, W. C. 1996. The art and practice of structure-based drug design: a molecular modeling perspective. Medicinal research reviews, 16(1): 3--50

work page 1996

[7] [7]

De Cao and T

Cao, N. D.; and Kipf, T. 2018. MolGAN: An implicit generative model for small molecular graphs. arXiv:1805.11973

work page arXiv 2018

[8] [8]

P.; Yu, G.; Herrington, D

Chang, Y.-T.; Hoffman, E. P.; Yu, G.; Herrington, D. M.; Clarke, R.; Wu, C.-T.; Chen, L.; and Wang, Y. 2019. Integrated identification of disease specific pathways using multi-omics data. bioRxiv, 666065

work page 2019

[9] [9]

Chen, J.; Hu, Y.; Wang, Y.; Lu, Y.; Cao, X.; Lin, M.; Xu, H.; Wu, J.; Xiao, C.; Sun, J.; et al. 2024 a . Trialbench: Multi-modal artificial intelligence-ready clinical trial datasets. arXiv preprint arXiv:2407.00631

work page arXiv 2024

[10] [10]

E.; Herrington, D

Chen, L.; Lu, Y.; Wu, C.-T.; Clarke, R.; Yu, G.; Van Eyk, J. E.; Herrington, D. M.; and Wang, Y. 2021. Data-driven detection of subtype-specific differentially expressed genes. Scientific reports, 11(1): 332

work page 2021

[11] [11]

Chen, T.; Hao, N.; Lu, Y.; and Van Rechem, C. 2024 b . Uncertainty Quantification on Clinical Trial Outcome Prediction. arXiv preprint arXiv:2401.03482

work page arXiv 2024

[12] [12]

V.; Chen, J.; and Fu, T

Chen, T.; Lu, Y.; Hao, N.; Rechem, C. V.; Chen, J.; and Fu, T. 2024 c . Uncertainty quantification and interpretability for clinical trial approval prediction. Health Data Science

work page 2024

[13] [13]

Delahaye, D.; Chaimatanan, S.; and Mongeau, M. 2019. Simulated annealing: From basics to applications. Handbook of metaheuristics, 1--35

work page 2019

[14] [14]

W.; and Sun, J

Fu, T.; Gao, W.; Coley, C. W.; and Sun, J. 2022 a . Reinforced Genetic Algorithm for Structure-based Drug Design. In Annual Conference on Neural Information Processing Systems (NeurIPS)

work page 2022

[15] [15]

W.; and Sun, J

Fu, T.; Gao, W.; Xiao, C.; Yasonik, J.; Coley, C. W.; and Sun, J. 2022 b . Differentiable Scaffolding Tree for Molecular Optimization. International Conference on Learning Representations

work page 2022

[16] [16]

M.; and Sun, J

Fu, T.; Xiao, C.; Li, X.; Glass, L. M.; and Sun, J. 2021. MIMOSA : Multi-constraint Molecule Sampling for Molecule Optimization. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 35, 125--133

work page 2021

[17] [17]

Gao, W.; Fu, T.; Sun, J.; and Coley, C. W. 2022. Sample Efficiency matters: benchmarking molecular optimization. Neural Information Processing Systems (NeurIPS) Track on Datasets and Benchmarks

work page 2022

[18] [18]

Gao, W.; Mercado, R.; and Coley, C. W. 2022 a . Amortized Tree Generation for Bottom-up Synthesis Planning and Synthesizable Molecular Design. arXiv:2110.06389

work page arXiv 2022

[19] [19]

Gao, W.; Mercado, R.; and Coley, C. W. 2022 b . Amortized Tree Generation for Bottom-up Synthesis Planning and Synthesizable Molecular Design. International Conference on Learning Representations

work page 2022

[20] [20]

M.; Fu, T.; Xiao, C.; and Sun, J

Glass, L. M.; Fu, T.; Xiao, C.; and Sun, J. 2021. MOLER: Incorporate molecule-level reward to enhance deep generative model for molecule optimization. IEEE transactions on knowledge and data engineering, 34(11): 5459--5471

work page 2021

[21] [21]

N.; Duvenaud, D.; Hern \'a ndez-Lobato, J

G \'o mez-Bombarelli, R.; Wei, J. N.; Duvenaud, D.; Hern \'a ndez-Lobato, J. M.; S \'a nchez-Lengeling, B.; Sheberla, D.; Aguilera-Iparraguirre, J.; Hirzel, T. D.; Adams, R. P.; and Aspuru-Guzik, A. 2018. Automatic chemical design using a data-driven continuous representation of molecules. ACS central science, 4(2): 268--276

work page 2018

[22] [22]

Objective-Reinforced Generative Adversarial Networks (ORGAN) for Sequence Generation Models

Guimaraes, G. L.; Sanchez-Lengeling, B.; Outeiral, C.; Farias, P. L. C.; and Aspuru-Guzik, A. 2017. Objective-reinforced generative adversarial networks (ORGAN) for sequence generation models. arXiv preprint arXiv:1705.10843

work page internal anchor Pith review Pith/arXiv arXiv 2017

[23] [23]

Huang, T

Huang, K.; Fu, T.; Gao, W.; Zhao, Y.; Roohani, Y.; Leskovec, J.; Coley, C. W.; Xiao, C.; Sun, J.; and Zitnik, M. 2021. Therapeutics Data Commons: Machine Learning Datasets and Tasks for Drug Discovery and Development. arXiv:2102.09548

work page arXiv 2021

[24] [24]

Jensen, J. H. 2019. A graph-based genetic algorithm and generative model/Monte Carlo tree search for the exploration of chemical space. Chemical science, 10(12): 3567--3572

work page 2019

[25] [25]

Jin, W.; Barzilay, R.; and Jaakkola, T. 2018. Junction tree variational autoencoder for molecular graph generation. ICML

work page 2018

[26] [26]

Jin, W.; Barzilay, R.; and Jaakkola, T. 2020. Multi-objective molecule generation using interpretable substructures. In International Conference on Machine Learning, 4849--4859. PMLR

work page 2020

[27] [27]

Korovina, K.; Xu, S.; Kandasamy, K.; Neiswanger, W.; Poczos, B.; Schneider, J.; and Xing, E. 2020. ChemBO : Bayesian optimization of small organic molecules with synthesizable recommendations. In International Conference on Artificial Intelligence and Statistics, 3393--3403. PMLR

work page 2020

[28] [28]

Li, Y.; Zhang, L.; and Liu, Z. 2018. Multi-Objective De Novo Drug Design with Conditional Graph Generative Model. arXiv:1801.07299

work page internal anchor Pith review Pith/arXiv arXiv 2018

[29] [29]

Liu, M.; Yan, K.; Oztekin, B.; and Ji, S. 2021. GraphEBM: Molecular graph generation with energy-based models. arXiv preprint arXiv:2102.00546

work page arXiv 2021

[30] [30]

Lu, Y.; and Liu, X.-Y. 2023. Reinforcement Learning for Ising Model. In Thirty-seventh Conference on Neural Information Processing Systems Track on Machine Learning for Physical Sciences

work page 2023

[31] [31]

J.; Cheng, Z.; Saylor, G.; Van Eyk, J

Lu, Y.; Wu, C.-T.; Parker, S. J.; Cheng, Z.; Saylor, G.; Van Eyk, J. E.; Yu, G.; Clarke, R.; Herrington, D. M.; and Wang, Y. 2022. COT : an efficient and accurate method for detecting marker genes among many subtypes. Bioinformatics Advances, 2(1): vbac037

work page 2022

[32] [32]

Luo, Y.; Yan, K.; and Ji, S. 2021. GraphDF : A discrete flow model for molecular graph generation. Proceedings of the 38th International Conference on Machine Learning, ICML , 139: 7192--7203

work page 2021

[33] [33]

Nigam, A.; Friederich, P.; Krenn, M.; and Aspuru-Guzik, A. 2020. Augmenting Genetic Algorithms with Deep Neural Networks for Exploring the Chemical Space. In The International Conference on Learning Representations (ICLR)

work page 2020

[34] [34]

Olivecrona, M.; Blaschke, T.; Engkvist, O.; and Chen, H. 2017 a . Molecular de-novo design through deep reinforcement learning. Journal of Cheminformatics

work page 2017

[35] [35]

Olivecrona, M.; Blaschke, T.; Engkvist, O.; and Chen, H. 2017 b . Molecular De Novo Design through Deep Reinforcement Learning. CoRR, abs/1704.07555

work page internal anchor Pith review Pith/arXiv arXiv 2017

[36] [36]

Rajak, A.; Suzuki, S.; Dutta, A.; and Chakrabarti, B. K. 2023. Quantum annealing: An overview. Philosophical Transactions of the Royal Society A, 381(2241): 20210417

work page 2023

[37] [37]

Shen, C.; Krenn, M.; Eppel, S.; and Aspuru-Guzik, A. 2021. Deep Molecular Dreaming: Inverse machine learning for de-novo molecular design and interpretability with surjective representations. Machine Learning: Science and Technology

work page 2021

[38] [38]

Shi, C.; Xu, M.; Zhu, Z.; Zhang, W.; Zhang, M.; and Tang, J. 2020. GraphAF : a Flow-based Autoregressive Model for Molecular Graph Generation. In The International Conference on Learning Representations (ICLR)

work page 2020

[39] [39]

Sterling, T.; and Irwin, J. J. 2015. ZINC 15--Ligand Discovery for Everyone. Journal of Chemical Information and Modeling, 55(11): 2324--2337

work page 2015

[40] [40]

Sun, J.; and Fu, T. 2022. Antibody complementarity determining regions (cdrs) design using constrained energy model. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, 389--399

work page 2022

[41] [41]

J.; Lu, Y.; Van Eyk, J

Wu, C.-T.; Shen, M.; Du, D.; Cheng, Z.; Parker, S. J.; Lu, Y.; Van Eyk, J. E.; Yu, G.; Clarke, R.; Herrington, D. M.; et al. 2022. Cosbin: cosine score-based iterative normalization of biologically diverse samples. Bioinformatics Advances, 2(1): vbac076

work page 2022

[42] [42]

You, J.; et al. 2018. Graph Convolutional Policy Network for Goal-directed Molecular Graph Generation. In Proceedings of the 32Nd International Conference on Neural Information Processing Systems, 6412--6422. Curran Associates Inc

work page 2018

[43] [43]

N.; and Riley, P

Zhou, Z.; Kearnes, S.; Li, L.; Zare, R. N.; and Riley, P. 2019. Optimization of molecules via deep reinforcement learning. Scientific reports

work page 2019