![[2602.18476] BioLM-Score: Language-Prior Conditioned Probabilistic Geometric Potentials for Protein-Ligand Scoring](https://arxiv.org/static/browse/0.3.4/images/arxiv-logo-fb.png){kind=logo}
[2602.18476] BioLM-Score: Language-Prior Conditioned Probabilistic Geometric Potentials for Protein-Ligand Scoring
Summary
BioLM-Score introduces a novel protein-ligand scoring model that enhances efficiency and interpretability in drug design by integrating geometric modeling with advanced representation learning.
Why It Matters
This research addresses the limitations of traditional scoring methods in drug discovery, offering a more efficient and generalizable approach. By improving the scoring process, BioLM-Score can significantly impact the speed and accuracy of drug development, which is crucial for addressing health challenges.
Key Takeaways
- BioLM-Score combines geometric modeling with representation learning for protein-ligand scoring.
- The model improves computational efficiency and generalization compared to traditional methods.
- It utilizes biomolecular language models to enhance structural and chemical representations.
- Evaluations show significant improvements in docking, scoring, and ranking tasks.
- BioLM-Score serves as an effective optimization objective for docking protocols.
Quantitative Biology > Biomolecules arXiv:2602.18476 (q-bio) [Submitted on 9 Feb 2026] Title:BioLM-Score: Language-Prior Conditioned Probabilistic Geometric Potentials for Protein-Ligand Scoring Authors:Zhangfan Yang, Baoyun Chen, Dong Xu, Jia Wang, Ruibin Bai, Junkai Ji, Zexuan Zhu View a PDF of the paper titled BioLM-Score: Language-Prior Conditioned Probabilistic Geometric Potentials for Protein-Ligand Scoring, by Zhangfan Yang and 6 other authors View PDF HTML (experimental) Abstract:Protein-ligand scoring is a central component of structure-based drug design, underpinning molecular docking, virtual screening, and pose optimization. Conventional physics-based energy functions are often computationally expensive, limiting their utility in large-scale screening. In contrast, deep learning-based scoring models offer improved computational efficiency but frequently suffer from limited cross-target generalization and poor interpretability, which restrict their practical applicability. Here we present BioLM-Score, a simple yet generalizable protein-ligand scoring model that couples geometric modeling with representation learning. Specifically, it employs modality-specific and structure-aware encoders for proteins and ligands, each augmented with biomolecular language models to enrich structural and chemical representations. Subsequently, these representations are integrated through a mixture density network to predict multimodal interatomic distance distributions, from which...
