PETA:Parameter-Efficient Test-Time Adaptation for Virtual Screening

Jia-Qi Lin; Yinghua Yao; Chang-Dong Wang; Yew-Soon Ong; Yuangang Pan

cs.LGAug 20, 2026

PETA:Parameter-Efficient Test-Time Adaptation for Virtual Screening

Authors: Jia-Qi Lin, Yinghua Yao, Chang-Dong Wang, Yew-Soon Ong, Yuangang Pan

Organizations: Centre for Frontier AI Research (CFAR), Agency for Science, Technology and Research (A*STAR), Singapore 138632 · School of Computer Science and Engineering, Sun Yat-sen University, Guangzhou, China

Abstract

Accurately ranking active ligands for a target protein pocket from massive chemical libraries remains a central challenge in virtual screening. DrugCLIP and its recent extensions substantially accelerate this process by encoding protein pockets and molecules into a shared embedding space. Despite this progress, further performance improvements typically require retraining the entire model, incurring substantial computational overhead and making target-specific customization inefficient. In this work, we formulate the specialization of pretrained virtual screening models to individual pockets as a test-time adaptation problem and propose PETA, a parameter-efficient framework that directly adapts pretrained model at test time. Given a target pocket, PETA constructs pocket-specific negatives through molecular diffusion and chemical validity filtering, and further moves them toward the reference ligand retrieved from structural databases via embedding-space mixup to create more challenging ranking tasks. A ranking objective then places greater emphasis on suppressing high-scoring invalid candidates that could contaminate the top-ranked screening results, providing structured supervision for lightweight adaptation. Experiments across diverse benchmarks demonstrate that this lightweight, pocket-specific adaptation outperforms both pretrained and fully retrained baselines while updating only the LayerNorm parameters, which account for approximately $0.03\%$ of the full model.

Figures & tables

Figure 1: PETA enables effective and parameter-efficient adaptation for virtual screening.

Figure 2: Framework of PETA. (a) Given a target pocket, PETA retrieves a reference ligand from a matched holo complex, constructs pocket-aware negatives, and adapts only the LayerNorm parameters of the ligand encoder. (b) The adapted model is then used for virtual screening over the test molecular library.

Figure 3: Distributions of per-pocket average pocket–ligand cosine similarities on the DUD-E virtual screening benchmark.

Figure 4: Reference ligand and pocket-conditioned candidates for ABL1. The crystal structure PDB 2HZI represents an experimentally resolved binding conformation of ABL1 in complex with the reference ligand PD180970. The generated candidates exhibit structural similarity to the reference ligand, providing pocket-relevant samples for adaptation.

Method	AUROC	BEDROC	EF 0.5%	EF 1%	EF 5%
Vina	71.60	–	9.13	7.32	4.44
Glide-SP	76.70	40.70	19.39	16.18	7.23
RFscore	65.21	12.41	4.90	4.52	2.98
Pafnucy	63.11	16.50	4.24	3.86	3.76
OnionNet	59.71	8.62	2.84	2.84	2.20
PLANET	71.60	–	10.23	8.83	5.40

Table 1: Virtual screening results on DUD-E. The best results are highlighted in bold .

Method	AUROC	BEDROC	EF 0.5%	EF 1%	EF 5%
Surflex	51.47	–	–	2.50	–
Glide-SP	53.15	4.00	3.17	3.41	2.01
PLANET	57.31	–	4.64	3.87	2.43
GNINA	60.93	5.40	–	4.63	–
DeepDTA	56.27	2.53	–	1.47	–
BigBind	60.80	–	–	3.82	–

Table 2: Virtual screening results on LIT-PCBA. The best results are highlighted in bold .

Panel of Figure 5 — Figure 5: Fine-grained ligand-ranking performance on the four-target FEP+ benchmark. Higher values indicate better performance.

$r_{\mathbf{x}^{u}}$	$\mathcal{G}_{\mathbf{x}^{u}}^{\mathrm{inv\textrm{-}e}}$	$\mathcal{G}_{\mathbf{x}^{u}}^{\mathrm{inv\textrm{-}h}}$	Mixup	AUROC	BEDROC	EF 0.5%	EF 1%	EF 5%
$\times$	$\checkmark$	$\checkmark$	$\checkmark$	80.16	48.11	38.49	31.50	10.64
$\checkmark$	$\times$	$\times$	$\checkmark$	81.23	51.74	40.41	31.90	10.80
$\checkmark$	$\times$	$\checkmark$	$\checkmark$	81.48	51.59	40.05	32.52	10.89
$\checkmark$	$\checkmark$	$\times$	$\checkmark$	81.23	51.80	40.55	32.82	11.02
$\checkmark$	$\checkmark$	$\checkmark$	$\times$	81.32	51.29	41.12	33.18	11.35
$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$	82.62	53.86	42.34	34.65	11.44

Table 3: Ablation studies on DUD-E. The best results are highlighted in bold .

$r_{\mathbf{x}^{u}}$	$\mathcal{G}_{\mathbf{x}^{u}}^{\mathrm{inv\textrm{-}e}}$	$\mathcal{G}_{\mathbf{x}^{u}}^{\mathrm{inv\textrm{-}h}}$	Mixup	AUROC	BEDROC	EF 0.5%	EF 1%	EF 5%
$\times$	$\checkmark$	$\checkmark$	$\checkmark$	55.60	7.15	8.07	5.32	2.29
$\checkmark$	$\times$	$\times$	$\checkmark$	57.66	7.51	9.08	6.62	2.86
$\checkmark$	$\times$	$\checkmark$	$\checkmark$	58.02	7.82	9.29	6.99	3.09
$\checkmark$	$\checkmark$	$\times$	$\checkmark$	59.44	7.69	9.17	6.53	3.06
$\checkmark$	$\checkmark$	$\checkmark$	$\times$	58.03	7.88	9.73	7.12	2.97
$\checkmark$	$\checkmark$	$\checkmark$	$\checkmark$	57.56	8.84	11.62	7.29	3.22

Table 4: Ablation studies on LIT-PCBA. The best results are highlighted in bold .

Panel of Figure 6 — Figure 6: Ablation of $\alpha$ on virtual screening performance.

Virtual Screening

Virtual screening uses computational methods to prioritize molecules for testing against biological targets or desired properties.

See all 20 papers

BindCLIP: A Unified Contrastive-Generative Representation Learning Framework for Virtual Screening
Feb 16, 2026Anjie Qiao, Zhen Wang, Yaliang Li +2Virtual Screening Molecular Representation Learning
Sun Yat-sen University, Guangzhou, China · Alibaba Group, Hangzhou, China
Adapting Boltz-2 with limited experimental activity data improves early enrichment in virtual screening
Sep 21, 2026Kairi Furui, Masahito OhueProtein-Ligand Binding Affinity Prediction Virtual Screening
Department of Computer Science, School of Computing, Institute of Science Tokyo, Yokohama 226-8501, Japan
TopU-LBVS: A Realistic Multi Target Benchmark for Ligand Based Virtual Screening
Sep 24, 2026Surbhi Kumar, Yuhe Zhou, Varun Shiralkar +2Benchmark Design Virtual Screening
UT Dallas, Mathematical Sciences Richardson, TX, USA

Feb 16, 2026J/K move · Enter open · S save

BindCLIP: A Unified Contrastive-Generative Representation Learning Framework for Virtual Screening

Anjie Qiao, Zhen Wang, Yaliang Li, Jiahua Rao+1

Sun Yat-sen University, Guangzhou, China · Alibaba Group, Hangzhou, China

Virtual Screening Molecular Representation Learning Structure-Based Drug Design Contrastive Learning

Open paper AskPDF arXiv

Virtual screening aims to efficiently identify active ligands from massive chemical libraries for a given target pocket. Recent CLIP-style models such as DrugCLIP enable scalable virtual screening by embedding pockets and ligands into a shared space. However, our analyses indicate that such representations can be insensitive to fine-grained binding interactions and may rely on shortcut correlations in training data, limiting their ability to rank ligands by true binding compatibility. To address these issues, we propose BindCLIP, a unified contrastive-generative representation learning framework for virtual screening. BindCLIP jointly trains pocket and ligand encoders using CLIP-style contrastive learning together with a pocket-conditioned diffusion objective for binding pose generation, so that pose-level supervision directly shapes the retrieval embedding space toward interaction-relevant features. To further mitigate shortcut reliance, we introduce hard-negative augmentation and a ligand-ligand anchoring regularizer that prevents representation collapse. Experiments on two public benchmarks demonstrate consistent improvements over strong baselines. BindCLIP achieves substantial gains on challenging out-of-distribution virtual screening and improves ligand-analogue ranking on the FEP+ benchmark. Together, these results indicate that integrating generative, pose-level supervision with contrastive learning yields more interaction-aware embeddings and improves generalization in realistic screening settings, bringing virtual screening closer to real-world applicability.

Abstract

Virtual Screening

Explore similar work

BindCLIP: A Unified Contrastive-Generative Representation Learning Framework for Virtual Screening

Adapting Boltz-2 with limited experimental activity data improves early enrichment in virtual screening

TopU-LBVS: A Realistic Multi Target Benchmark for Ligand Based Virtual Screening