Representation-aware conformal prediction for reliable uncertainty quantification in drug-target affinity

Abstract

Motivation: Quantifying predictive uncertainty is essential for computational drug discovery to reduce experimental costs and false positives. However, most drug-target affinity (DTA) models focus on improving prediction accuracy without reliable uncertainty estimates.

Results: We propose representation-aware conformal prediction (RCP)-DTA, a representation-aware conformal prediction framework that constructs adaptive prediction intervals based on feature similarity. By constructing local calibration sets, our approach generates sample-specific prediction intervals that achieve empirical coverage close to the nominal confidence level. To improve representation quality, RCP-DTA employs a cross-modal architecture that integrates pretrained sequence representations and structural information. Binding site context is incorporated via virtual pocket feedback to enhance residue representations, followed by gated fusion for adaptive integration of sequence and structural modalities. Evaluations across multiple DTA benchmark datasets demonstrate that RCP-DTA achieves competitive affinity prediction performance while providing reliable uncertainty quantification.

Availability and implementation: Code and data are available at: https://github.com/Lee515628/RCP-DTA

Neural representation of the subjective uncertainty of prediction in the human brain
Xiaoyue Sun, Xizi Gong, Huan Luo +2
Communications Biology30 Sep 2026Peer-reviewedCC BY-NC-ND 4.0Full textNeuroscienceCrossref
Quantifying Uncertainty in Brain Age Predictions via Conformal Prediction
U. Samanta, G. Shafiei, H. T. Karim +2
bioRxiv4 Sep 2026PreprintCC BYFull textNeurosciencebioRxiv
Uncertainty Quantification-Incorporated Ensemble Model for Personalized Prediction of Severe Radiotherapy-Induced Immunosuppression Among Esophageal Cancer Patients
Baode Gao, Yan Chu, Zongsheng Hu +10
arXiv21 Sep 2026PreprintarXiv licenceFull textBioinformatics & computational biologyarXiv
Multicenter Assessment of Combined Brain and Cervical Spinal cord 3T MP2RAGE T1 Measurements for Reliable Tissue Microstructure Quantification
Malo Gaubert, Benoit Combès, Henitsoa Rasoanandrianina +9
NMR in Biomedicine8 Jul 2026Peer-reviewedCC BY 4.0Full textNeuroscienceHAL
Representation-aware calibration enables statistically reliable protein structure search
L. Shi, H. Zhou, Y. Yang +1
bioRxiv3 Nov 2026PreprintCC BYFull textBioinformatics & computational biologybioRxiv

Cite

BibTeX RIS

BibTeX

@article{li2026representationaware,
  title = {{Representation-aware conformal prediction for reliable uncertainty quantification in drug-target affinity}},
  author = {Xinya Li and Yaxin Zhang and Ximin Zeng and Danping Zhou and Yue Cheng and Shaoping Shi},
  journal = {Bioinformatics},
  year = {2026},
  month = oct,
  doi = {10.1093/bioinformatics/btag771},
  url = {https://academic.oup.com/bioinformatics/article/doi/10.1093/bioinformatics/btag771/8914808},
  note = {CC BY 4.0},
}

RIS

TY  - JOUR
TI  - Representation-aware conformal prediction for reliable uncertainty quantification in drug-target affinity
AU  - Xinya Li
AU  - Yaxin Zhang
AU  - Ximin Zeng
AU  - Danping Zhou
AU  - Yue Cheng
AU  - Shaoping Shi
PY  - 2026
DA  - 2026/10/13
JO  - Bioinformatics
DO  - 10.1093/bioinformatics/btag771
UR  - https://academic.oup.com/bioinformatics/article/doi/10.1093/bioinformatics/btag771/8914808
AB  - Motivation: Quantifying predictive uncertainty is essential for computational drug discovery to reduce experimental costs and false positives. However, most drug-target affinity (DTA) models focus on improving prediction accuracy without reliable uncertainty estimates.  Results: We propose representation-aware conformal prediction (RCP)-DTA, a representation-aware conformal prediction framework that constructs adaptive prediction intervals based on feature similarity. By constructing local calibration sets, our approach generates sample-specific prediction intervals that achieve empirical coverage close to the nominal confidence level. To improve representation quality, RCP-DTA employs a cross-modal architecture that integrates pretrained sequence representations and structural information. Binding site context is incorporated via virtual pocket feedback to enhance residue representations, followed by gated fusion for adaptive integration of sequence and structural modalities. Evaluations across multiple DTA benchmark datasets demonstrate that RCP-DTA achieves competitive affinity prediction performance while providing reliable uncertainty quantification.  Availability and implementation: Code and data are available at: https://github.com/Lee515628/RCP-DTA
SN  - 1367-4803
SN  - 1367-4811
ER  -

Abstract

Related

Cite