Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

End-to-End Reverse Screening Identifies Protein Targets of Small Molecules Using HelixFold3

Shengjie Xu, Xianbin Ye, Mengran Zhu, Xiaonan Zhang, Shanzhuo Zhang, Xiaomin Fang

TL;DR

The paper tackles reverse screening by eliminating error propagation through an end-to-end framework that jointly folds proteins and docks ligands using HelixFold3. This approach yields superior target identification and binding-site localization, demonstrated on ~100 ligands against ~1,000 candidate proteins, with Top-1% and Top-10% success rates of 38.0% and 71.0%, respectively. Retrospective case studies on Imatinib (drug repurposing) and Ibrutinib (off-target prediction) show robust prioritization of known targets and physiologically relevant off-targets, highlighting practical utility for mechanism elucidation and safety assessment. The framework provides a scalable, data-driven platform for rational drug discovery, though limitations with transmembrane targets and generalization signal avenues for future improvement.

Abstract

Identifying protein targets for small molecules, or reverse screening, is essential for understanding drug action, guiding compound repurposing, predicting off-target effects, and elucidating the molecular mechanisms of bioactive compounds. Despite its critical role, reverse screening remains challenging because accurately capturing interactions between a small molecule and structurally diverse proteins is inherently complex, and conventional step-wise workflows often propagate errors across decoupled steps such as target structure modeling, pocket identification, docking, and scoring. Here, we present an end-to-end reverse screening strategy leveraging HelixFold3, a high-accuracy biomolecular structure prediction model akin to AlphaFold3, which simultaneously models the folding of proteins from a protein library and the docking of small-molecule ligands within a unified framework. We validate this approach on a diverse and representative set of approximately one hundred small molecules. Compared with conventional reverse docking, our method improves screening accuracy and demonstrates enhanced structural fidelity, binding-site precision, and target prioritization. By systematically linking small molecules to their protein targets, this framework establishes a scalable and straightforward platform for dissecting molecular mechanisms, exploring off-target interactions, and supporting rational drug discovery.

End-to-End Reverse Screening Identifies Protein Targets of Small Molecules Using HelixFold3

TL;DR

The paper tackles reverse screening by eliminating error propagation through an end-to-end framework that jointly folds proteins and docks ligands using HelixFold3. This approach yields superior target identification and binding-site localization, demonstrated on ~100 ligands against ~1,000 candidate proteins, with Top-1% and Top-10% success rates of 38.0% and 71.0%, respectively. Retrospective case studies on Imatinib (drug repurposing) and Ibrutinib (off-target prediction) show robust prioritization of known targets and physiologically relevant off-targets, highlighting practical utility for mechanism elucidation and safety assessment. The framework provides a scalable, data-driven platform for rational drug discovery, though limitations with transmembrane targets and generalization signal avenues for future improvement.

Abstract

Identifying protein targets for small molecules, or reverse screening, is essential for understanding drug action, guiding compound repurposing, predicting off-target effects, and elucidating the molecular mechanisms of bioactive compounds. Despite its critical role, reverse screening remains challenging because accurately capturing interactions between a small molecule and structurally diverse proteins is inherently complex, and conventional step-wise workflows often propagate errors across decoupled steps such as target structure modeling, pocket identification, docking, and scoring. Here, we present an end-to-end reverse screening strategy leveraging HelixFold3, a high-accuracy biomolecular structure prediction model akin to AlphaFold3, which simultaneously models the folding of proteins from a protein library and the docking of small-molecule ligands within a unified framework. We validate this approach on a diverse and representative set of approximately one hundred small molecules. Compared with conventional reverse docking, our method improves screening accuracy and demonstrates enhanced structural fidelity, binding-site precision, and target prioritization. By systematically linking small molecules to their protein targets, this framework establishes a scalable and straightforward platform for dissecting molecular mechanisms, exploring off-target interactions, and supporting rational drug discovery.
Paper Structure (15 sections, 1 equation, 4 figures)

This paper contains 15 sections, 1 equation, 4 figures.

Table of Contents

  1. Introduction
  2. Results
  3. Benchmarking the End-to-End Reverse Screening Paradigm
  4. Determinants of Superior Screening Performance
  5. Retrospective Validation in Drug Repurposing
  6. Retrospective Investigation of Off-Target Prediction
  7. Discussion
  8. Methods
  9. Evaluation Datasets
  10. Negative Sample Selection
  11. Positive Sample Collection
  12. Final Benchmark Dataset
  13. Reverse Screening Pipelines
  14. Target Library Construction for Real-World Applications
  15. Evaluation Metrics

Figures (4)

  • Figure 1: Computational paradigms and applications of reverse screening.(a,b) Concept and therapeutic applications of reverse screening. (c) Conventional step-wise pipeline, highlighting the accumulation and propagation of errors across decoupled steps. (d) HelixFold3 end-to-end strategy, which unifies protein folding and ligand scoring into a single, direct predictive process.
  • Figure 2: Comprehensive performance evaluation of reverse screening methods.(a) Success rate comparison between step-wise method and our end-to-end method. (b) Impact of scoring function for the end-to-end method. (c) Structural prediction precision of the whole chain or pocket of proteins. The excellent, moderate, and distant groups are classified by the structure difference between predicted and crystallized protein structures (holo or apo). (d) Binding site localization precision comparing GT + pocket detection, predicted bounded (holo), and predicted unbounded (apo).
  • Figure 3: Application framework and case studies for reverse screening.(a) Imatinib case background: timeline and therapeutic targets (GIST: KIT/PDGFRA; CML: ABL1). (b) Ibrutinib case background: on-target anti-tumor efficacy (BTK) and off-target adverse events (bleeding: ITK, TXK, BLK; infection: JAK family; atrial fibrillation: KCNH2). (c) Imatinib case study: ranking comparison between ipTM and pLDDT metrics. (d) Ibrutinib case study: ranking comparison between ipTM and pLDDT metrics.
  • Figure 4: Benchmark dataset construction pipeline. Representative negative protein targets selected from AFDB (left) and positive protein-ligand pairs selected from recent PDB protein-ligand complexes (right).