Enhancing Virtual Screening of Cystathionine β-Synthase Inhibitors: Benchmarking Target-Specific Machine-Learning Scoring Functions Against State-of-the-Art AI Docking and Co-Folding Approaches

Cystathionine beta-synthase (CBS) has emerged as an important therapeutic target implicated in cancer and Down syndrome, yet the discovery of selective CBS inhibitors remains challenging due to limited structural diversity of known ligands and the scarcity of target-focused virtual screening (VS) benchmarks. In this study, we present the first comprehensive evaluation of CBS-specific machine-learning (ML) models for structure-based VS, supported by a carefully curated and up-to-date data set of experimentally validated CBS inhibitors, true inactives and decoys. Importantly, we benchmarked their performance against a diverse panel of 16 established VS pipelines, including classical docking-based scoring schemes, modern deep-learning (DL) docking tools, and recent co-folding approaches for protein-ligand modeling and affinity prediction. The CBS-specific ML classifiers substantially outperformed these state-of-the-art (SOTA) methods in early enrichment, highlighting the advantage of target-focused training for VS tasks. Our results further reveal that generic DL docking and co-folding approaches struggle to achieve reliable screening performance when applied to targets that are under-represented in or entirely absent from their training data, as appears to be the case for CBS. This underscores a key limitation of broadly trained foundation-style models in prospective drug discovery campaigns involving less-studied proteins.胱硫醚β合酶（Cystathionine beta-synthase, CBS）已成为癌症与唐氏综合征相关的关键治疗靶点，但由于已知配体的结构多样性有限，且缺乏靶点聚焦型虚拟筛选（virtual screening, VS）基准集，开发选择性CBS抑制剂仍极具挑战。 本研究首次针对基于结构的虚拟筛选，开展了CBS专属机器学习（machine-learning, ML）模型的全面评估，所用数据集为一套经精心整理且更新至最新的、经实验验证的CBS抑制剂、真实无活性化合物及诱饵分子数据集。值得注意的是，我们针对16套成熟的虚拟筛选流程组合开展了性能基准测试，涵盖经典的基于分子对接的评分方案、现代深度学习（deep-learning, DL）对接工具，以及新近用于蛋白质-配体建模与亲和力预测的共折叠方法。 CBS专属机器学习分类器在早期富集性能上显著优于这些当前最优（state-of-the-art, SOTA）方法，凸显了靶点聚焦型训练在虚拟筛选任务中的优势。我们的研究结果进一步揭示，通用型深度学习对接与共折叠方法，当应用于训练数据中占比极低或完全未覆盖的靶点时，难以获得可靠的筛选性能，而CBS恰好属于这类靶点。这一发现凸显了经过广泛训练的基础类模型，在涉及研究较少的蛋白质的前瞻性药物发现项目中存在的关键局限性。