Chemical and geometric representation fidelity improves drug--target affinity prediction

Abstract

Predicting drug--target binding affinity (DTA) requires models to distinguish subtle chemical and structural determinants underlying molecular recognition. Although recent approaches increasingly incorporate richer drug and protein information, such information may be compressed, homogenized or discretized during representation construction, causing affinity-relevant distinctions to be lost before interaction modelling. We hypothesized that this representation-stage information loss constitutes an upstream bottleneck that cannot be reliably overcome by increasingly complex interaction predictors. To test this hypothesis, we developed ReGeoDTA, a representation-preserving framework that maintains affinity-relevant chemical heterogeneity in molecular representations and continuous geometric relationships in protein structures. Across three benchmark datasets, ReGeoDTA consistently improved affinity prediction, and the proposed representation-preserving strategies retained their benefits across diverse DTA architectures. Controlled representation degradation progressively reduced predictive performance, whereas increasing downstream predictor complexity failed to recover information lost during representation construction. These findings identify representation fidelity as an upstream design principle for accurate and generalizable drug--target affinity prediction, with potential implications for computational compound prioritization.

Topology-aware few-shot segmentation improves rice leaf-cell boundary and count fidelity, while correction is risky under tested domain shifts
Gaixing Wei, Da Zeng, Shenglin Qin +1
Frontiers in Plant Science13 Oct 2026Peer-reviewedCC BY 4.0Full textBiophysicsCrossrefPMC
BRIDGE-KO: auditable transfer learning with bridged knockoffs improves cross-cohort microbiome prediction of immunotherapy response
Briefings in bioinformatics1 Sep 2026Peer-reviewedCC BYFull textMolecular biology & gene regulationEurope PMC
Representation learning for low-data protein fitness prediction using sequence–structure embeddings and graph neural networks
Chanoknun Phunnasorn, Narupat Hongdilokkul, Wasakorn Laesanklang +2
Scientific Reports16 Oct 2026Peer-reviewedCC BY-NC-ND 4.0Full textMolecular biology & gene regulationCrossref
Multimodal alignment improves generalizability of genomic biomarker prediction in computational pathology
Ekaterina Redekop, Eric Zimmermann, Ava P. Amini +5
Cell Reports Methods8 Sep 2026Peer-reviewedCC BY-NC-ND 4.0Full textMolecular biology & gene regulationPMC
ProteinJEPA: Latent prediction improves protein language model pretraining
Dan Ofer, Dafna Shahaf, Michal Linial
arXiv23 Sep 2026PreprintCC BY-NC-SA 4.0Full textBiophysics StatisticsarXiv

Cite

BibTeX RIS

BibTeX

@misc{li2026chemical,
  title = {{Chemical and geometric representation fidelity improves drug--target affinity prediction}},
  author = {Yixiao Li and Yining Qian and Yefan Chen and Zenghui Chen and Jiayue Sun and Yuhai Zhao and Cheng Tan and An-Yang Lu},
  howpublished = {arXiv},
  year = {2026},
  month = sep,
  doi = {10.48550/arxiv.2609.13230},
  eprint = {2609.13230},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2609.13230},
}

RIS

TY  - UNPB
TI  - Chemical and geometric representation fidelity improves drug--target affinity prediction
AU  - Yixiao Li
AU  - Yining Qian
AU  - Yefan Chen
AU  - Zenghui Chen
AU  - Jiayue Sun
AU  - Yuhai Zhao
AU  - Cheng Tan
AU  - An-Yang Lu
PY  - 2026
DA  - 2026/09/02
PB  - arXiv
DO  - 10.48550/arxiv.2609.13230
UR  - https://arxiv.org/abs/2609.13230
AB  - Predicting drug--target binding affinity (DTA) requires models to distinguish subtle chemical and structural determinants underlying molecular recognition. Although recent approaches increasingly incorporate richer drug and protein information, such information may be compressed, homogenized or discretized during representation construction, causing affinity-relevant distinctions to be lost before interaction modelling. We hypothesized that this representation-stage information loss constitutes an upstream bottleneck that cannot be reliably overcome by increasingly complex interaction predictors. To test this hypothesis, we developed ReGeoDTA, a representation-preserving framework that maintains affinity-relevant chemical heterogeneity in molecular representations and continuous geometric relationships in protein structures. Across three benchmark datasets, ReGeoDTA consistently improved affinity prediction, and the proposed representation-preserving strategies retained their benefits across diverse DTA architectures. Controlled representation degradation progressively reduced predictive performance, whereas increasing downstream predictor complexity failed to recover information lost during representation construction. These findings identify representation fidelity as an upstream design principle for accurate and generalizable drug--target affinity prediction, with potential implications for computational compound prioritization.
ER  -

Abstract

Related

Cite