Skip to content
DraconiticResearch
Search Sources API
203,948 studies31 sourcesUpdated 22 min ago
PreprintCC BY-NCbioRxiv

ReTIF: Granularity-Aware Multitask Interaction Routing for RNA-Compound Interaction Prediction and Binding-Site Localization

  • A. Chang
  • H. Zhu
  • H. Chen
  • C. Wang
  • X. Wang
  • X. Zeng
  • Y. Ding
  • P. Xiong
  • S. K. Zhou1
  1. 1University of Science and Technology of China
Read full textPDFDOICite

Abstract

RNA-targeted drug discovery requires both RNA-compound interaction prediction and nucleotide-level binding-site (BS) localization. These two tasks rely on different levels of interaction information: DTI prediction summarizes overall RNA-compound compatibility, whereas BS localization requires preserving nucleotide-level compound-associated signals. However, existing multitask models often use shared cross-modal representations for both tasks before task-specific prediction layers, which may limit their ability to preserve task-dependent interaction patterns. We therefore propose ReTIF (Relation-enhanced Task-specific Interaction Framework), which constructs separate cross-modal interaction representations for DTI prediction and BS localization before aggregation. ReTIF integrates multi-source RNA-compound representations from frozen RNA-FM, StructRFM, Mole-BERT, and MolFormer encoders, and builds separate interaction representations for DTI scoring and BS localization. The DTI branch captures global compatibility through aggregation, whereas the BS branch preserves nucleotide-compound resolution and enhances local evidence through relation-guided propagation with RNA structural and compound topological priors. An asymmetric DTI-derived compatibility signal provides global context to BS prediction while maintaining local evidence. Across five-fold evaluations under unseen pair, RNA, compound, and joint shifts with 13 baselines, ReTIF has the highest mean in 13 of 16 scenario-metric combinations, including BS AUPR in all four settings.

Related

  1. Protein Language Model-Conditioned Graph Neural Networks for Multitask GPCR Ligand Activity Prediction
    M. De, E. Lodh, S. Majumder +1
    bioRxivPreprintAll rights reservedFull textBioinformatics & computational biologybioRxiv
  2. A biological-response compound representation allows chemical perturbation prediction across cell lines
    L. Le Breton, L. Kaufmann, E. Carraz-Billat +2
    bioRxivPreprintCC BYFull textBioinformatics & computational biologybioRxiv
  3. Physical interaction with Ephrin B1 promotes CXCR4 intracellular localization and oncogenic potential
    Alessandro Rabbito, Omolade Otun, Amos Fumagalli +12
    Cellular and Molecular Life SciencesPeer-reviewedCC BY-NC-ND 4.0Full textProteomics & metabolomicsCell biologyHAL
  4. A method for brain tumor detection and localization by means of deep learning and reliable prediction explainability
    Marcello Di Giammarco, Antonella Santone, Mario Cesarelli +2
    PLOS OnePeer-reviewedCC BYFull textNeuroscienceMedical imaging & radiologyPLOSEurope PMCPMC
  5. Repurposing PeTriBERT for Protein Protein Interaction Prediction with Sequence Structure Fusion
    Arthur Wavreille, Gabriel Krouk, Baldwin Dumortier
    HALPreprintCC BY 4.0Full textBioinformatics & computational biologyHALbioRxiv

Cite

BibTeXRIS
BibTeX
@misc{chang2026retif,
  title = {{ReTIF: Granularity-Aware Multitask Interaction Routing for RNA-Compound Interaction Prediction and Binding-Site Localization}},
  author = {A. Chang and H. Zhu and H. Chen and C. Wang and X. Wang and X. Zeng and Y. Ding and P. Xiong and S. K. Zhou},
  howpublished = {bioRxiv},
  year = {2026},
  month = sep,
  doi = {10.64898/2026.09.02.748859},
  url = {https://www.biorxiv.org/content/10.64898/2026.09.02.748859v1},
  note = {CC BY-NC},
}
RIS
TY  - UNPB
TI  - ReTIF: Granularity-Aware Multitask Interaction Routing for RNA-Compound Interaction Prediction and Binding-Site Localization
AU  - A. Chang
AU  - H. Zhu
AU  - H. Chen
AU  - C. Wang
AU  - X. Wang
AU  - X. Zeng
AU  - Y. Ding
AU  - P. Xiong
AU  - S. K. Zhou
PY  - 2026
DA  - 2026/09/02
PB  - bioRxiv
DO  - 10.64898/2026.09.02.748859
UR  - https://www.biorxiv.org/content/10.64898/2026.09.02.748859v1
AB  - RNA-targeted drug discovery requires both RNA-compound interaction prediction and nucleotide-level binding-site (BS) localization. These two tasks rely on different levels of interaction information: DTI prediction summarizes overall RNA-compound compatibility, whereas BS localization requires preserving nucleotide-level compound-associated signals. However, existing multitask models often use shared cross-modal representations for both tasks before task-specific prediction layers, which may limit their ability to preserve task-dependent interaction patterns. We therefore propose ReTIF (Relation-enhanced Task-specific Interaction Framework), which constructs separate cross-modal interaction representations for DTI prediction and BS localization before aggregation. ReTIF integrates multi-source RNA-compound representations from frozen RNA-FM, StructRFM, Mole-BERT, and MolFormer encoders, and builds separate interaction representations for DTI scoring and BS localization. The DTI branch captures global compatibility through aggregation, whereas the BS branch preserves nucleotide-compound resolution and enhances local evidence through relation-guided propagation with RNA structural and compound topological priors. An asymmetric DTI-derived compatibility signal provides global context to BS prediction while maintaining local evidence. Across five-fold evaluations under unseen pair, RNA, compound, and joint shifts with 13 baselines, ReTIF has the highest mean in 13 of 16 scenario-metric combinations, including BS AUPR in all four settings.
ER  -