Arman SalehiContact

Biomedical AI · Published

Comparative study of hybrid quantum-classical models for cold-start drug–target affinity prediction: a partial pilot analysis

Arman Salehi, Ghazal Salehi, Ashkan Heydarian, Hamid Reza Goudarzi

PaperDOIExplore model ↓Code artifact
Venue
Scientific Reports
Published
2026-09-01
DOI
10.1038/s41598-026-69754-2
Citations
Citation indexing pending · checked 2026-09-01
Last verified
2026-09-01

Research question

Can a hybrid quantum–classical multimodal model predict binding affinity when the drug, target, or both were unseen during training?

Condensed abstract

The study compares QVGAT-DPI with a specification-based implementation of Q-BAFNet under cold-start drug–target affinity protocols. QVGAT-DPI combines molecular graphs, BRICS fragments, molecular fingerprints, ChemBERTa drug representations, full-sequence ESM-2 protein representations, bidirectional cross-attention, and classically simulated residual variational quantum modules.

Key contribution

A residual hybrid architecture that adds quantum-derived corrections to classical attention and topology paths, evaluated with entity-level protocols for unseen drugs (S2), unseen targets (S3), and jointly unseen drugs and targets (S4).

Method overview

Drug inputs branch through graph-attention layers, an unordered BRICS-fragment set, a Morgan fingerprint, and ChemBERTa. Overlapping ESM-2 windows represent complete protein sequences. Bidirectional cross-attention models drug–protein interactions; six-qubit state-vector circuits supply gated residuals to attention and topology features before sample-specific fusion and regression.

QVGAT-DPI model journey

How one pair becomes an affinity estimate

Follow the drug and protein through five stages. Select any module for its role, or run the complete explanatory trace.

Illustrative inputs and motion · topology follows the published model

Cold-start protocol
Path visibility
Pair

Cold-start status changes the evaluation split, not the input format. The example identities below are deliberately illustrative.

Stage 1 of 5

Cold-start evidence atlas

Select a dataset, protocol, and endpoint. Complete cells reveal paired fold differences; partial and unrun cells stay blank.

Complete · 10/10 paired folds
Dataset
Protocol
Endpoint
Completed paired folds by dataset and cold-start protocol
DatasetS2 · drugS3 · targetS4 · both
DAVIS
KIBA
DAVIS S2 RMSE paired-fold differencesTen fold-level RMSE differences defined as QVGAT-DPI minus Q-BAFNet. Lower values favor QVGAT-DPI.favors QVGAT-DPIfavors Q-BAFNetF0Fold 0: −0.142552−0.142552F1Fold 1: −0.129284−0.129284F2Fold 2: −0.259584−0.259584F3Fold 3: −0.100561−0.100561F4Fold 4: +0.023380+0.023380F5Fold 5: −0.130010−0.130010F6Fold 6: +0.010593+0.010593F7Fold 7: −0.047671−0.047671F8Fold 8: −0.031854−0.031854F9Fold 9: −0.133469−0.133469mean−0.7000000+0.150000
Paired difference summary

Δ = QVGAT-DPI − Q-BAFNet · lower RMSE / higher CI is favorable

Mean Δ
−0.094101
Bootstrap 95% interval
−0.144692 to −0.045941
Exact p
0.00976562
Holm-adjusted p
0.0391

The average paired change favors QVGAT-DPI for RMSE. This was the only tested contrast that remained below 0.05 after Holm correction.

S2 exact paired-fold records
DAVIS S2, QVGAT-DPI and Q-BAFNet fold results
FoldQVGAT CIQ-BAFNet CIΔ CIQVGAT RMSEQ-BAFNet RMSEΔ RMSE
00.7102420.562881+0.1473620.7016210.844173−0.142552
10.6926040.788627−0.0960231.1055881.234872−0.129284
20.6877490.701508−0.0137590.7842251.043809−0.259584
30.7434950.769435−0.0259400.5524210.652981−0.100561
40.6997810.668143+0.0316380.8382880.814908+0.023380
50.8101850.622598+0.1875860.5817900.711800−0.130010
60.6497060.665326−0.0156190.9639130.953320+0.010593
70.5371790.522140+0.0150390.8542420.901912−0.047671
80.6584220.485214+0.1732090.8756840.907538−0.031854
90.7462650.548541+0.1977250.9147711.048240−0.133469
All complete-protocol statistical summaries
DAVIS paired-difference summaries reported in Appendix E
ProtocolEndpointMean ΔBootstrap 95% intervalExact pHolm p
S2CI+0.060122−0.000480 to +0.1234410.099609380.19921875
S2RMSE−0.094101−0.144692 to −0.0459410.009765620.03906250
S4CI+0.087226+0.032326 to +0.1427500.019531250.05859375
S4RMSE−0.135053−0.294289 to −0.0050120.119140620.19921875
Original explanatory rendering of the published method. The drug, protein, prediction trace, and moving signals are illustrative; no interaction-level prediction is reproduced. Circuits were evaluated with state-vector simulation, not a QPU. Fold values are the paired Appendix E pilot records, and incomplete cells remain uninterpreted.

Dataset and evaluation

Data

The pilot uses the DAVIS and KIBA drug–target affinity benchmarks. Only DAVIS S2 and S4 produced complete ten-fold paired result sets; DAVIS S3 and all KIBA protocol cells remained partial or unrun.

Protocol

The planned comparison used ten deterministic folds per cold-start protocol. Completed, fold-matched QVGAT-DPI and paper-specification Q-BAFNet results were summarized with paired differences, bootstrap confidence intervals, exact p-values, and Holm correction across the primary CI and RMSE endpoints.

Main results

For complete DAVIS S2, the mean QVGAT-DPI minus Q-BAFNet difference was +0.060122 for CI and −0.094101 for RMSE; only the RMSE result remained below 0.05 after Holm correction (p = 0.0391). DAVIS S4 favored QVGAT-DPI in mean CI and RMSE differences, but neither endpoint remained below 0.05 after correction. Neither complete protocol met the prespecified dual-endpoint superiority rule.

DAVIS S2 · Δ CI+0.060122QVGAT-DPI − Q-BAFNet; Holm p = 0.1992
DAVIS S2 · Δ RMSE−0.094101Lower is favorable; Holm p = 0.0391
DAVIS S4 · Δ CI+0.087226QVGAT-DPI − Q-BAFNet; Holm p = 0.0586
DAVIS S4 · Δ RMSE−0.135053Lower is favorable; Holm p = 0.1992

Baseline comparison

Against the local paper-specification Q-BAFNet comparator, QVGAT-DPI had favorable mean differences on both primary endpoints in complete DAVIS S2 and S4. Multiplicity-adjusted support was limited to S2 RMSE. Q-BAFNet is a hybrid comparator, not a classical baseline or an exact reproduction claim.

Limitations and failure modes

  • All quantum circuits were state-vector simulations; no quantum processing unit was used, and the study does not establish quantum advantage.
  • The experimental grid was incomplete: only DAVIS S2 and S4 reached ten paired folds, so partial DAVIS S3 and KIBA runs are not confirmatory evidence.
  • The reported comparison does not isolate a causal contribution from the residual quantum modules.
  • Benchmark affinity prediction does not establish prospective biological, experimental, or clinical utility.

Reproducibility resources

The paper-associated computational material is available as a versioned Zenodo artifact.

Verification sources

Cite this work

Download .bibDownload .ris

Arman Salehi, Ghazal Salehi, Ashkan Heydarian, Hamid Reza Goudarzi (2026). Comparative study of hybrid quantum-classical models for cold-start drug–target affinity prediction: a partial pilot analysis. Scientific Reports. https://doi.org/10.1038/s41598-026-69754-2

BibTeX
@article{salehi2026qvgat,
  title = {Comparative study of hybrid quantum-classical models for cold-start drug–target affinity prediction: a partial pilot analysis},
  author = {Arman Salehi and Ghazal Salehi and Ashkan Heydarian and Hamid Reza Goudarzi},
  year = {2026},
  doi = {10.1038/s41598-026-69754-2},
  journal = {Scientific Reports},
  publisher = {Springer Nature}
}

Publication record last verified 2026-09-01.