Quantum Machine Learning for Predicting Binding Free Energies in Structure-Based Virtual Screening

Pei-Kun Yang

REVIEW 5 major objections 5 minor 37 references

Quantum Machine Learning for Predicting Binding Free Energies in Structure-Based Virtual Screening

T0 review · 5 major / 5 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 nine-qubit parameterized quantum circuit, trained on voxelized protein–ligand structures, predicts binding free energies with an RMSD of 2.37 kcal/mol and Pearson correlation of 0.650 on the PDBbind Core set, and keeps ranking intact…

desk verdict A plausible but under-validated QML feasibility study: the headline RMSD/PCC is selected on the test set, and without classical baselines the quantum component isn't actually shown to be doing the work. read the letter →

arxiv 2507.18425 v1 pith:RMQILCT2 submitted 2025-07-24 quant-ph

Pei-Kun Yang This is my paper · ORCID

classification quant-ph

keywords quantummachinelearningbindingfreeenergypredictionstructure-basedvirtualscreeningparameterizedcircuitsprotein-ligandvoxelencodingPDBbindNISQrobustness

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper proposes that a small parameterized quantum circuit can serve as a structure-based virtual screening scoring function: nine qubits encode a coarse three-dimensional occupancy grid of a protein–ligand complex, and the measurement of a single qubit is converted directly into a predicted binding free energy. The author trains this circuit on 19,158 complexes and reports that six circuit units predict the held-out Core set with an RMSD of 2.37 kcal/mol and a Pearson correlation of 0.650. The ranking of ligands remains stable under finite-shot sampling and under simulated amplitude-damping and depolarizing noise, which the paper reads as evidence that ranking-based screening can run on near-term quantum hardware. The paper also sketches how ancillary index qubits could let one circuit process many complexes in parallel, a route toward screening very large chemical libraries.

What carries the argument

The central object is the QCbind circuit built from repeated QMLunits, where each unit contains an Lpar layer of parameterized RX and RZ single-qubit rotations followed by an Lbreaker layer of CNOT gates with a distinct entanglement topology per qubit index. Nine qubits are used: three encode atom-type information and six encode spatial positions in a discretized 4×4×4 voxel grid derived from 512 occupancy values. Instead of measuring all qubits, only the 0th qubit is measured, and the predicted binding free energy is read out as $\Delta G = 100(y_0-y_1)$, where $y_0$ and $y_1$ are the probabilities of measuring $|0\rangle$ and $|1\rangle$. Training minimizes the mean squared error between this readout and experimental free energies, and a parallel extension uses $m$ ancillary index qubits to process $2^m$ complexes through a block-diagonal unitary.

What would settle it

Retrain the same six-unit circuit after randomly permuting the 512 voxel occupancies across the training complexes; if the held-out RMSD and Pearson correlation stay near 2.37 kcal/mol and 0.650, the reported accuracy does not come from the structural encoding.

Watch

Extended reading notes

Core claim

The central claim is that a relatively shallow nine-qubit circuit—six stacked QMLunits, each combining single-qubit rotations with CNOT entanglement layers—can learn a transferable binding-energy scoring function from voxelized structures. On the PDBbind Core test set, the trained circuit reaches an RMSD of 2.37 kcal/mol and a Pearson correlation of 0.650, with accuracy improving up to six units and declining beyond. Sampling 100,000 measurement shots from a single qubit keeps predictions close to the ideal full-state results, with a maximum Pearson correlation of 0.632. Under simulated amplitude damping and depolarizing noise, the absolute error worsens to about 2.97 kcal/mol but the correlation holds at 0.653, which the paper interprets as preservation of ligand ranking rather than exact energy estimation.

Load-bearing premise

The load-bearing premise is that reading out one qubit's measurement probabilities after only nine qubits encode a 512-voxel occupancy vector leaves enough chemical information to reproduce the experimental binding free energy ($\Delta G$).

Editorial extensions

If this is right

Five to six QMLunits are the useful circuit depth: predictive accuracy improves up to six units and declines afterward, so near-term quantum circuits do not need to be deep to be useful.
A single-qubit readout with about 100,000 measurement shots closely reproduces full-state predictions, substantially lowering the measurement overhead required for inference.
Under the tested noise levels, absolute prediction errors grow but the Pearson correlation remains around 0.65, meaning ligand ranking can still be trusted even when absolute energies are degraded.
Adding ancillary index qubits allows one circuit to evaluate many complexes at once through vectorized block-diagonal matrix operations, offering a concrete scaling path for high-throughput virtual screening.

Reading between the lines

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

The paper does not compare the quantum model against classical voxel-based scoring functions on the same split; a head-to-head benchmark would clarify whether the quantum circuit adds predictive power or merely matches an encoding-equivalent classical model.
Because the readout uses only one qubit, a natural testable extension is to measure additional qubits or learn the readout observable, which may improve absolute accuracy without requiring deeper circuits.
The noise robustness is demonstrated at two fixed low rates; scaling the damping and depolarizing rates would reveal how much noise the ranking guarantee can tolerate before it breaks.
The parallel-screening argument assumes ideal hardware and an efficient classical-to-quantum encoding, so practical advantage over classical screening remains a projection rather than a demonstrated outcome.

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.

Desk Editor's Note

read the letter

Colleague,

Quick read: this is a legitimate application of an existing PQC building block (the QMLunit from Zhao et al. 2021) to the PDBbind binding-free-energy task, packaged with a clean noise/sampling robustness study and a GitHub code link. The interesting result is not the absolute numbers; it's that the ranking metric (PCC) stays roughly constant when you add shot noise or light depolarizing/amplitude-damping noise. That is a useful sanity check for anyone thinking about near-term QML scoring.

What is new is modest: the architecture is borrowed, the voxel encoding is straight out of classical 3D-CNN scoring, and the task is standard. The contribution is the specific 9-qubit readout design and the three-mode evaluation. Fine as a feasibility note.

The soft spots are real but not fatal.

First, the depth of six QMLunits is chosen after looking at the test-set curves. The text says 'the test-set performance was with six QMLunits' and that beyond it metrics declined. That makes the 2.37 kcal/mol / 0.650 a post-selection maximum, not an unbiased estimate. You need a validation split or nested CV, plus variance across the random initializations. Without that, 'six beats five' is possibly seed noise.

Second, there is no classical baseline trained on the same 512 occupancy features. A linear or MLP model on those same features would anchor whether the quantum circuit is adding anything. Without it, the headline could be the features doing all the work.

Third, the parallel estimation section is not right. Putting m ancilla qubits around a 9-qubit circuit and writing a block-diagonal matrix gives you a superposition of outputs, not simultaneous classical answers; measuring collapses to one. Also the 'U29*29' notation looks like a typo for 2^9. This section should be corrected or removed.

Minor: the intro's claim that 200 more qubits could evaluate 10^60 complexes is hand-waving. The single-qubit readout is a limitation worth noting but not disqualifying: one real number per complex is weak, but the experiment is what it is.

Who is this for? People working on QML for drug screening who want a quick feasibility data point. It deserves serious peer review because the topic is timely and the code/data are promised, but it needs major revision before publication: baselines, a proper model-selection protocol, error bars, and a fixed parallelism section.

Referee Report

5 major / 5 minor

Summary. The paper introduces a parameterized quantum circuit model, built from repeated QMLunits, to predict protein–ligand binding free energies in structure-based virtual screening. Protein–ligand complexes from PDBbind v2020 are voxelized into 512 occupancy values that are encoded into a nine-qubit circuit; the model is trained with mean squared error and evaluated on the PDBbind Core set under full simulation, finite-shot sampling, and noisy simulation. The main claim is that with six QMLunits the model achieves an RMSD of 2.37 kcal/mol and a Pearson correlation of 0.650 on the test set, with relative ranking largely preserved under noise. The paper also proposes a parallel estimation scheme using ancillary qubits to process multiple complexes simultaneously.

Significance. If the central empirical result is reliable, the paper would demonstrate that a compact nine-qubit parameterized circuit can learn a transferable binding-affinity scoring function from structural features, which is a useful step toward NISQ-era virtual screening. The manuscript is relatively self-contained in its data-preparation and simulation methodology, and it ships a GitHub repository with the code. The strengths are the use of an external benchmark (PDBbind) with a standard train/test split and the evaluation under three realistic inference conditions. However, the significance is currently limited by an optimistic test-set-based model selection, lack of statistical error bars, absence of any classical baseline on the same features, an underspecified feature-to-qubit encoding, and a conceptually flawed parallel-estimation argument.

major comments (5)

[Results, 'Estimation of ΔGbind Using Full Probability'; Fig. 4; Methods, 'Training Protocol'] The headline result (six QMLunits, RMSD 2.37 kcal/mol, PCC 0.650) is selected after evaluating depths 1–10 on the test set, as shown in Fig. 4 and stated in the text ('The test-set performance was with six QMLunits'). The Methods, however, state that the retained model is the one with the lowest training-set RMSD, which does not define a rule for choosing the depth. Because the depth is chosen on the same data used to report the final metric, the reported numbers are post-selection maxima and are optimistically biased estimates of generalization. Please either pre-register a validation-based depth-selection rule (e.g., using a separate validation split) or report the test-set metrics for all depths along with the selection procedure, and discuss the resulting optimistic bias.
[Results (Fig. 3) and Methods, 'Training Protocol'] No statistical uncertainty is reported for any metric. The Methods state that three independent training runs are performed, but only the single run with the lowest training RMSD is retained, and no variance across seeds is shown. Consequently, differences between adjacent depths (e.g., five vs. six QMLunits) could be within seed noise. Please report means and standard deviations over the random initializations for the test-set RMSD and PCC, and justify that the reported depth is stable across seeds.
[Methods, 'Parallel Estimation of Multiple Protein–Ligand Complexes'; Eqs. (4)–(5); Fig. 5] The central claim of this section—that the outputs O^j for all 2^m inputs can be 'computed simultaneously' using a block-diagonal unitary—is not valid as a quantum algorithm. In quantum mechanics, applying a block-diagonal U to a superposition of index states entangles the index register with the data register; a single measurement collapses the state and reveals only one bitstring, not all 2^m outputs. To recover all outputs, one would need to repeat the experiment many times and incur measurement overhead, so the factorization in Eq. (5) is a classical vectorized matrix operation, not a demonstration of quantum parallel prediction. This directly affects the scalability motivation in the Introduction and Conclusions. Please either correct the claim to describe a classical batch-processing trick or provide a concrete quantum measurement scheme that avoids the collapse issue.
[Methods, 'Quantum Circuit Architecture' and 'Training Protocol'] The manuscript never specifies how the 512 occupancy values are encoded into the initial state of the nine-qubit circuit. The text only says the 'normalized occupancy vector was encoded into a quantum state' and names three type qubits and six position qubits, but the mapping from 512 floats to the initial amplitudes or rotation angles is absent. This is a load-bearing technical detail: without it the architecture is not reproducible from the manuscript, and it is impossible to tell whether the predictive power comes from the circuit, the encoding, or the readout. Please state the encoding explicitly (e.g., angle encoding, amplitude encoding, or a learned classical preprocessing layer) or provide the precise circuit definition used in the code.
[Results and Discussion (throughout)] No classical baseline is trained or evaluated on the same 512 occupancy features. As a result, the reported RMSD of 2.37 kcal/mol and PCC of 0.650 cannot be attributed to the quantum circuit specifically; a classical model (e.g., ridge regression, random forest, or a feedforward network) on the same voxel features might achieve comparable or better accuracy. Adding such a baseline is necessary to support the paper's stated motivation that the quantum approach offers a practical alternative for virtual screening. Without it, the claims about quantum advantage or even quantum competitiveness are not yet established.

minor comments (5)

[Abstract and Results] The Abstract states 'even when using 100,000 shots' while the Results section also reports 10,000 and 1,000,000 shots. Please align the abstract with the full set of shot counts or explicitly state that 100,000 is the representative setting.
[Fig. 3 caption] The caption lists colors (black, red, orange, blue) but the legend in the figure itself is hard to distinguish in the printed version; consider adding markers or line styles and enlarge the font.
[Eq. (4) and surrounding text] The notation 'U29∗29' is unclear; it should be written as a 2^9 × 2^9 unitary (or 512 × 512) with proper subscripts, and the dimensions of the identity matrices in Eq. (4) should be specified.
[Methods, 'Dataset Preparation for Training and Validation'] Reference 37 is cited for the Core subset, but the appropriate citation is the CASF benchmark papers or the original PDBbind follow-up descriptions; the current reference is about deep learning in virtual screening and does not describe the Core set.
[Introduction, last paragraph] The statement 'adding just 200 more qubits could enable the simultaneous evaluation of up to 10^60 complexes' is overstated and depends on the parallel-estimation scheme criticized above; please temper this claim or remove it until a valid quantum measurement protocol is provided.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the central result is an empirical benchmark trained on external PDBbind labels and evaluated on a held-out Core set; no prediction reduces to its inputs by construction.

full rationale

The paper's central claim is an empirical machine-learning benchmark, not a first-principles derivation: the model is trained on the PDBbind v2020 training set (19,158 complexes) and evaluated on the held-out Core subset (285 complexes). The predicted binding free energy is read out through Eq. (3), ΔG = 100(y0-y1), where y0-y1 is a trained expectation value; the factor 100 is a fixed scaling choice and is not fitted to the test labels. The reported RMSD and PCC therefore compare externally measured experimental affinities against circuit outputs, and the result could have failed. No parameter is fitted to the Core set: the Methods state that "the model with the lowest RMSD on the training set was selected for evaluation on the test set." The choice of six QMLunits is, however, made after inspecting test-set performance across depths 1-10, which introduces optimistic selection bias rather than circularity; the reported 2.37 kcal/mol and 0.650 are post-selection numbers, but they are not equivalent to the model's inputs by construction. The architecture is said to follow "the framework introduced by Zhao et al. (Zhao, 2021)," but that reference is missing from the reference list; this is a missing-citation concern, not a circularity, because the architecture is a design choice and does not by itself justify the accuracy claim. There is no self-citation chain, no imported uniqueness theorem, and no renamed known result. The derivation chain is self-contained with respect to circularity: the prediction is an independently evaluated empirical outcome.

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

The method depends on hand-set structural encoding choices, a linear single-qubit readout, and the test-set-selected circuit depth. No new physical entities are introduced. The scalability claim relies on an unproven assumption about preparing and reading out superposed inputs.

free parameters (4)

Output scaling factor (100 in Eq. 3) = 100 (hand-set)
Maps the probability difference of qubit 0 to kcal/mol; set manually, not optimized or derived.
Voxel encoding constants = 16 Å grid, 32^3 grid, max-pool factor 8, vdW radii C/N/O/other 1.9/1.8/1.7/2.0 Å
Input features depend on these constants from prior docking literature; no sensitivity analysis provided.
Occupancy normalization target = 0.5 per protein and ligand
Chosen to balance quantum state encoding; no test of alternative normalizations.
Number of QMLunits = 5-6 units (best at 6)
Selected based on test-set performance, not an independent validation set; this is a hyperparameter tuned on the evaluation benchmark.

assumptions (5)

domain assumption Binding free energies from PDBbind v2020 are treated as reliable experimental ground truth.
Used as training labels and test targets; errors in these measurements propagate directly into reported RMSD/PCC.
domain assumption Voxelized occupancy grid (32^3, max-pooled to 4^3 per channel, normalized to 0.5 sum of squares per protein/ligand) preserves enough structural and chemical information to predict binding affinity.
The entire input representation rests on this encoding; no ablation or comparison to alternative encodings is given.
ad hoc to paper A 9-qubit PQC with single-qubit readout and linear scaling dG = 100*(y0-y1) can represent the structure-to-affinity relationship.
This architectural assumption is not derived or supported by universality arguments; it is validated only empirically on one dataset split.
ad hoc to paper For parallel estimation, one can prepare 2^m input states in superposition and read out all outputs without measurement collapse (Eq. 4-5).
The block-diagonal matrix description is a classical batch computation; the paper does not provide a measurement scheme that avoids collapse, so this premise is unsupported.
domain assumption Amplitude damping (rate 0.001) and depolarizing noise (p=0.0005) after each gate approximate near-term hardware errors.
Chosen rather than measured on a specific device; conclusions about hardware compatibility follow only if this noise model is representative.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Quantum Machine Learning for Predicting Binding Free Energies in Structure-Based Virtual Screening." pith.science (2026). https://pith.science/paper/RMQILCT2

@misc{pith2026250718425,
  author       = {Pith},
  title        = {Pith review of: Quantum Machine Learning for Predicting Binding Free Energies in Structure-Based Virtual Screening},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RMQILCT2}},
  note         = {Machine review of arXiv:2507.18425}
}

read the original abstract

In structure-based virtual screening, it is often necessary to evaluate the binding free energy of protein-ligand complexes by considering not only molecular conformations but also how these structures shift and rotate in space. The number of possible combinations grows rapidly and can become overwhelming. While classical computing has limitations in this context, quantum computing offers a promising alternative due to its inherent parallelism. In this study, we introduce a quantum machine learning approach that encodes molecular information into quantum states and processes them using parameterized quantum gates. The model is implemented and trained using PyTorch, and its performance is evaluated under three settings: ideal simulation, limited-shot sampling, and simulations with quantum noise. With six quantum circuit units, the model achieves an RMSD of 2.37 kcal/mol and a Pearson correlation of 0.650. Even when using 100,000 shots, the predictions remain consistent, indicating that the model is compatible with near-term quantum hardware. Although noise slightly reduces accuracy, the ranking of ligand affinities remains largely unchanged. These findings point to a practical and scalable strategy that balances robustness and predictive power, offering a viable path to accelerate virtual screening through moderately deep quantum circuits.

Figures

Figures reproduced from arXiv: 2507.18425 by the authors.

**Figure 1.** (a) Overall quantum circuit architecture, consisting of multiple stacked QMLunits. Each QMLunit includes two consecutive layers: a parameterized single-qubit rotation layer (Lpar) and an entanglement layer (Lbreaker). (b) Internal structure of the Lpar layer, where each qubit is acted on by a pair of RX and RZ rotation gates, each with its own trainable parameter. (a) Q2(T0) Q3(Z1) Q4(Z0) Q5(Y1) Q6(Y0) Q7(X1) Q0(T2)… view at source ↗

**Figure 2.** (a) Overview of the entanglement module Lbreaker, designed to introduce controlled entanglement across qubits. (b–j) Configurations of Lbreaker(0) through Lbreaker(8), where each circuit consists of two layers of CNOT gates. In each Lbreaker(n), the n th qubit is explicitly excluded from being a target. Filled dots indicate control qubits, and symbols indicate target qubits. Training Protocol. The quantum circuit mo… view at source ↗

**Figure 4.** Scatter plots of predicted versus experimental Gbind on the test set under fullprobability inference for circuits with varying numbers of QMLunits. Each panel corresponds to a different QMLunit depth, ranging from 1 to 10. Predicted values are generated by the trained quantum model and compared against experimental values. The (a) (b) (c) (d) (e) (f) (g) (h) (i) (j) [PITH_FULL_IMAGE:figures/full_fig_p010_4.png] view at source ↗

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

37 extracted references · 34 canonical work pages

[1]

H. M. Berman et al., The protein data bank. Nucleic acids research 28, 235-242 (2000)

work page 2000
[2]

E. H. B. Maia, L. C. Assis, T. A. De Oliveira, A. M. Da Silva, A. G. Taranto, Structure-based virtual screening: from classical to artificial intelligence. Frontiers in chemistry 8, 343 (2020)

work page 2020
[3]

Fu et al., Accurate determination of protein: ligand standard binding free energies from molecular dynamics simulations

H. Fu et al., Accurate determination of protein: ligand standard binding free energies from molecular dynamics simulations. Nature Protocols 17, 1114-1141 (2022)

work page 2022
[4]

C. R. Groom, I. J. Bruno, M. P. Lightfoot, S. C. Ward, The Cambridge structural database. Acta Crystallographica Section B: Structural Science, Crystal Engineering and Materials 72, 171-179 (2016)

work page 2016
[5]

Ruddigkeit, R

L. Ruddigkeit, R. Van Deursen, L. C. Blum, J.-L. Reymond, Enumeration of 166 billion organic small molecules in the chemical universe database GDB-17. Journal of chemical information and modeling 52, 2864-2875 (2012)

work page 2012
[6]

C. Yang, Y . Zhang, Lin_F9: A Linear Empirical Scoring Function for Protein– Ligand Docking. Journal of Chemical Information and Modeling, (2021)

work page 2021
[7]

Crampon, A

K. Crampon, A. Giorkallos, M. Deldossi, S. Baud, L. A. Steffenel, Machine- learning methods for ligand–protein molecular docking. Drug discovery today, (2021)

work page 2021
[8]

Shen et al., Can machine learning consistently improve the scoring power of classical scoring functions? Insights into the role of machine learning in scoring functions

C. Shen et al., Can machine learning consistently improve the scoring power of classical scoring functions? Insights into the role of machine learning in scoring functions. Briefings in bioinformatics 22, 497-514 (2021)

work page 2021

Show all 37 references

[9]

H. Li, K. H. Sze, G. Lu, P. J. Ballester, Machine‐learning scoring functions for structure‐based virtual screening. Wiley Interdisciplinary Reviews: Computational Molecular Science 11, e1478 (2021). 15

2021
[10]

Kadukova, K

M. Kadukova, K. d. S. Machado, P. Chacón, S. Grudinin, KORP-PL: a coarse- grained knowledge-based scoring function for protein–ligand interactions. Bioinformatics 37, 943-950 (2021)

2021
[11]

Gaillard, Evaluation of AutoDock and AutoDock Vina on the CASF-2013 benchmark

T. Gaillard, Evaluation of AutoDock and AutoDock Vina on the CASF-2013 benchmark. Journal of chemical information and modeling 58, 1697-1706 (2018)

2018
[12]

Yasuo, M

N. Yasuo, M. Sekijima, Improved method of structure-based virtual screening via interaction-energy-based learning. Journal of chemical information and modeling 59, 1050-1061 (2019)

2019
[13]

Ragoza, J

M. Ragoza, J. Hochuli, E. Idrobo, J. Sunseri, D. R. Koes, Protein–ligand scoring with convolutional neural networks. Journal of chemical information and modeling 57, 942-957 (2017)

2017
[14]

A. T. McNutt et al., GNINA 1.0: molecular docking with deep learning. Journal of cheminformatics 13, 1-20 (2021)

2021
[15]

J. J. Irwin et al., ZINC20—a free ultralarge-scale chemical database for ligand discovery. Journal of chemical information and modeling 60, 6065-6073 (2020)

2020
[16]

Reymond, The chemical space project

J.-L. Reymond, The chemical space project. Accounts of Chemical Research 48, 722-730 (2015)

2015
[17]

Reymond, R

J.-L. Reymond, R. Van Deursen, L. C. Blum, L. Ruddigkeit, Chemical space as a source for new drugs. MedChemComm 1, 30-38 (2010)

2010
[18]

de Lima Marquezino, R

F. de Lima Marquezino, R. Portugal, C. Lavor, A primer on uantum computing. (Springer, 2019)

2019
[19]

Brown, Z

S. Brown, Z. Vranesic, EBOOK: Fundamentals of Digital Logic. (McGraw Hill, 2008)

2008
[20]

Alvarez-Rodriguez, M

U. Alvarez-Rodriguez, M. Sanz, L. Lamata, E. Solano, Quantum artificial life in an IBM quantum computer. Scientific reports 8, 1-9 (2018)

2018
[21]

Arute et al., Quantum supremacy using a programmable superconducting processor

F. Arute et al., Quantum supremacy using a programmable superconducting processor. Nature 574, 505-510 (2019)

2019
[22]

Zhong et al., Quantum computational advantage using photons

H.-S. Zhong et al., Quantum computational advantage using photons. Science 370, 1460-1463 (2020)

2020
[23]

Wright et al., Benchmarking an 11-qubit quantum computer

K. Wright et al., Benchmarking an 11-qubit quantum computer. Nature communications 10, 1-6 (2019)

2019
[24]

Blinov, B

S. Blinov, B. Wu, C. Monroe, Comparison of cloud-based ion trap and superconducting quantum computer architectures. AVS Quantum Science 3, 033801 (2021)

2021
[25]

J. Li, R. O. Topaloglu, S. Ghosh, Quantum generative models for small molecule drug discovery. IEEE Transactions on Quantum Engineering 2, 1-8 (2021). 16

2021
[26]

Batra et al., Quantum Machine Learning Algorithms for Drug Discovery Applications

K. Batra et al., Quantum Machine Learning Algorithms for Drug Discovery Applications. (2021)

2021
[27]

I. Cong, S. Choi, M. D. Lukin, Quantum convolutional neural networks. Nature Physics 15, 1273-1278 (2019)

2019
[28]

MacCormack, C

I. MacCormack, C. Delaney, A. Galda, N. Aggarwal, P. Narang, Branching quantum convolutional neural networks. Physical Review Research 4, 013117 (2022)

2022
[29]

Chen, Quantum Dilated Convolutional Neural Networks

Y . Chen, Quantum Dilated Convolutional Neural Networks. IEEE Access, (2022)

2022
[30]

S. Wei, Y . Chen, Z. Zhou, G. Long, A Quantum Convolutional Neural Network on NISQ Devices. arXiv preprint arXiv:2104.06918, (2021)

2021 arXiv
[31]

F. Shen, J. Liu, QFCNN: Quantum Fourier Convolutional Neural Network. arXiv preprint arXiv:2106.10421, (2021)

2021 arXiv
[32]

Liu et al., Hybrid quantum-classical convolutional neural networks

J. Liu et al., Hybrid quantum-classical convolutional neural networks. Science China Physics, Mechanics & Astronomy 64, 1-8 (2021)

2021
[33]

Jing et al., RGB Image Classification with Quantum Convolutional Ansaetze

Y . Jing et al., RGB Image Classification with Quantum Convolutional Ansaetze. arXiv preprint arXiv:2107.11099, (2021)

2021 arXiv
[34]

T. Hur, L. Kim, D. K. Park, Quantum convolutional neural network for classical data classification. arXiv preprint arXiv:2108.00661, (2021)

2021 arXiv
[35]

Kerenidis, J

I. Kerenidis, J. Landman, A. Prakash, Quantum algorithms for deep convolutional neural networks. arXiv preprint arXiv:1911.01117, (2019)

2019 arXiv
[36]

R. Wang, X. Fang, Y . Lu, C.-Y . Yang, S. Wang, The PDBbind database: methodologies and updates. Journal of medicinal chemistry 48, 4111-4119 (2005)

2005
[37]

T. B. Kimber, Y . Chen, A. V olkamer, Deep Learning in Virtual Screening: Recent Applications and Developments. International Journal of Molecular Sciences 22, 4435 (2021)

2021

Pith tools

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

[1] [1]

H. M. Berman et al., The protein data bank. Nucleic acids research 28, 235-242 (2000)

work page 2000

[2] [2]

E. H. B. Maia, L. C. Assis, T. A. De Oliveira, A. M. Da Silva, A. G. Taranto, Structure-based virtual screening: from classical to artificial intelligence. Frontiers in chemistry 8, 343 (2020)

work page 2020

[3] [3]

Fu et al., Accurate determination of protein: ligand standard binding free energies from molecular dynamics simulations

H. Fu et al., Accurate determination of protein: ligand standard binding free energies from molecular dynamics simulations. Nature Protocols 17, 1114-1141 (2022)

work page 2022

[4] [4]

C. R. Groom, I. J. Bruno, M. P. Lightfoot, S. C. Ward, The Cambridge structural database. Acta Crystallographica Section B: Structural Science, Crystal Engineering and Materials 72, 171-179 (2016)

work page 2016

[5] [5]

Ruddigkeit, R

L. Ruddigkeit, R. Van Deursen, L. C. Blum, J.-L. Reymond, Enumeration of 166 billion organic small molecules in the chemical universe database GDB-17. Journal of chemical information and modeling 52, 2864-2875 (2012)

work page 2012

[6] [6]

C. Yang, Y . Zhang, Lin_F9: A Linear Empirical Scoring Function for Protein– Ligand Docking. Journal of Chemical Information and Modeling, (2021)

work page 2021

[7] [7]

Crampon, A

K. Crampon, A. Giorkallos, M. Deldossi, S. Baud, L. A. Steffenel, Machine- learning methods for ligand–protein molecular docking. Drug discovery today, (2021)

work page 2021

[8] [8]

Shen et al., Can machine learning consistently improve the scoring power of classical scoring functions? Insights into the role of machine learning in scoring functions

C. Shen et al., Can machine learning consistently improve the scoring power of classical scoring functions? Insights into the role of machine learning in scoring functions. Briefings in bioinformatics 22, 497-514 (2021)

work page 2021