2026-10-14T11:38:31Z https://api.figshare.com/v2/oai

oai:figshare.com:article/33857771 2026-09-16T15:15:34Z category_1 category_4 category_7 category_13 category_14 category_19 category_21 category_873 category_734 category_931 category_132 portal_63 item_type_3 month_year_09_2026

HelixDTA: Dual-Branch Sequence–Structure Learning with Complete Target Structures for Robust and Interpretable Drug–Target Affinity Prediction Shang Lou (3444725) Xuhua Li (4844508) Yujie Peng (7064474) Yuan Yuan (25702) Bo Peng (273834) Yingyu Yi (24992171) Xuchao Zhang (396841) Hailong Zhao (8774726) Beilei Wang (298371) Kun Li (277035) Hong Qian (135570) Tao Ren (72849) Hongcang Gu (130793) Fan Zhang (46132) Biophysics Biochemistry Medicine Genetics Molecular Biology Pharmacology Biotechnology Chemical Sciences not elsewhere classified Biological Sciences not elsewhere classified Information Systems not elsewhere classified Infectious Diseases protein pocket structures novel protein targets machine learning methods independently learns representations highlighting molecular regions governs molecular recognition despite recent advances complete target structures level structural context integrating complete target demonstrates significant potential based pose inspection architecture preserves modality integrating structural context demonstrating robust generalization helixdta achieved strong either sequence information drug discovery pipeline sequence context integrating sequence potential use branch architecture based cold physical information drug discovery drug candidates derived information unseen scaffolds two distinct thus failing start settings specific insights robust solution prediction layer precision demands limitation prevents kiba benchmarks highly accurate fully capturing derived features critical three complementary modalities competitive performance affinity prediction Accurate prediction of drug–target affinity (DTA) is crucial for accelerating the discovery of drug candidates. Despite recent advances in machine learning methods, existing models often rely solely on either sequence information or protein pocket structures. This limitation prevents these models from fully capturing the critical three-dimensional (3D) physical information that governs molecular recognition, thus failing to meet the high-precision demands of drug discovery. To address this, we introduce HelixDTA, a deep learning framework featuring a parallel dual-branch architecture that independently learns representations from two distinct but complementary modalities: sequence context and complete target structures. By integrating sequence-derived features with structure-derived features at the prediction layer, this architecture preserves modality-specific insights while characterizing drug–target interactions more comprehensively. HelixDTA achieved strong and competitive performance on the Davis and KIBA benchmarks and maintained clear advantages across most similarity-based cold-start settings, demonstrating robust generalization to compounds with unseen scaffolds and novel protein targets. Furthermore, its built-in attention mechanism enhances interpretability by highlighting molecular regions that contribute to affinity prediction. Finally, a WRN case study demonstrates the potential use of HelixDTA for structure-aware allosteric candidate prioritization by integrating complete target-level structural context with docking-based pose inspection. In conclusion, HelixDTA offers a highly accurate, interpretable, and robust solution for DTA prediction. It highlights the value of integrating structural context with sequence-derived information and demonstrates significant potential to empower precision drug design and accelerate the drug discovery pipeline. 2026-09-16T00:00:00Z Dataset Dataset 10.1021/acs.jcim.6c01759.s001 https://figshare.com/articles/dataset/HelixDTA_Dual-Branch_Sequence_Structure_Learning_with_Complete_Target_Structures_for_Robust_and_Interpretable_Drug_Target_Affinity_Prediction/33857771 CC BY-NC 4.0