PMRD separates mechanism-consistent factors from modality-specific information and constructs a consensus response domain across three modalities and combines complementary representations through reliability-aware multiview retrieval and supports PMRD as an effective framework for mechanism-aware multimodal drug representation learning.
Abstract
Multimodal drug discovery enables drug representation learning beyond chemical structure by incorporating cellular responses such as gene expression and cell morphology. However, direct fusion and instance-level contrastive alignment may mix mechanism-related signals with modality-specific noise and incorrectly separate structurally dissimilar but biologically related compounds. This limitation can obscure transferable mechanism patterns required for predicting the properties of unseen compounds. We introduce PMRD, a pharmacological response domain-guided framework for multimodal zero-shot drug property prediction. PMRD separates mechanism-consistent factors from modality-specific information and constructs a consensus response domain across three modalities. Mechanism candidate augmentation identifies locally stable factors, while retrieval-geometry attribution dynamically reweights the alignment and augmentation objectives according to whether their updates preserve inter-drug discriminability.This feedback suppresses training signals that conflict with mechanism-discriminative retrieval. PMRD further combines complementary representations through reliability-aware multiview retrieval. Experiments on public datasets show improved zero-shot property prediction and more biologically coherent drug neighborhoods. Hard-negative analysis further indicates fewer conflicts between structurally dissimilar but response-related compounds. These results support PMRD as an effective framework for mechanism-aware multimodal drug representation learning.\footnote{The code will be released upon publication.}
Abstract Motivation Accurate prediction of drug-target interactions (DTIs) is fundamental to drug discovery and mechanistic understanding. While deep learning has advanced computational DTI prediction, most existing methods rely primarily on molecular structural representations, including drug structures and protein sequences, while overlooking cellular phenotypes that reflect downstream biological effects. Cell Painting enables high-content morphological profiling that captures systems-level responses to chemical and genetic perturbations but remains underutilized in DTI modeling. Integrating molecular information with cellular phenotypes offers an opportunity to improve both predictive performance and biological interpretability. Results We propose a two-stage contrastive learning framework integrating drug structures, protein sequences, and Cell Painting morphological profiles into a unified embedding space. Stage 1 learns modality-specific representations independently from structure-based and image-based data; Stage 2 aligns these via multi-positive contrastive learning to bridge molecular structural information with cellular phenotypes. Cross-modal retrieval achieves median Recall@10 values of 0.77 (random split) and 0.33 (scaffold split), outperforming bilinear and random baselines. In external DTI prediction on the BIOSNAP dataset, our model achieves an AUC of 0.92 with image-based representations and 0.90 under structure-only settings, surpassing existing methods. Model interpretation via integrated gradients reveals pathway-specific morphological signatures associated with drug targets, providing biologically interpretable insights into drug mechanisms. Availability https://github.com/YJRubyLai/Unified-DTI
Predicting anticancer drug synergy is pivotal for personalizing combination therapies; however, existing deep learning models often rely heavily on complex, high-dimensional multi-omics data and precomputed molecular properties. Such dependence increases data acquisition barriers and limits model applicability in resource-constrained or rapid screening scenarios. In this study, we propose DDI-MMAF, a lightweight cross-modal framework that avoids using explicit high-dimensional omics profiles as direct model inputs. It utilizes a minimalist input protocol consisting of drug SMILES sequences and verbalized biological context, encompassing cell line names and their corresponding tissue origins. The architecture integrates a domain-specific semantic encoder to extract coarse-grained biomedical semantic priors from cell-line nomenclature and a deep residual visual network to capture hierarchical spatial features from molecular images. The core innovation lies in a multi-modal affine fusion mechanism that dynamically modulates molecular visual features conditioned on biological semantic embeddings. Systematic evaluations demonstrate that despite its simplified inputs, the model achieves a ROC AUC of 0.934 on benchmark datasets, showing the best performance against methods that utilize explicit omics information or handcrafted molecular descriptors. Furthermore, our approach maintains robust performance under the internal scaffold-split setting, while also achieving competitive performance compared with the evaluated baselines on the independent AstraZeneca blind test set. Overall, this research demonstrates that effective semantic-guided modulation enables accurate synergy prediction from raw minimalist inputs, offering a practical and efficient computational solution for cost-effective drug combination discovery.
Hao Li, Qianhui Jiang, Jiahui Guan et al.· European journal of medicina...· 0 citations
Abstract Motivation Accurate prediction of synergistic drug combinations can accelerate anticancer combination discovery. Existing methods inadequately model higher order drug–drug–cell-line interactions and drug–disease associations and remain sensitive to sparse and noisy multiomics data, limiting generalization to unseen cell lines and drug combinations. Results We present Momentum Contrast (MoCo)-MultiSynergy, a multimodal framework that combines modality-specific momentum contrastive learning with heterogeneous hypergraph modeling. The hypergraph represents synergistic drug–drug–cell-line triplets and drug–disease associations, while gated residual propagation refines node representations. MoCo modules regularize encoded drug and cell-line representations using latent feature masking and Gaussian perturbation. On the O’Neil and NCI-ALMANAC datasets, MoCo-MultiSynergy achieves the highest AUROC and AUPRC across the evaluated settings, with the largest gains when generalizing to unseen cell lines and drug combinations. Availability and implementation Source code is available at https://github.com/27167199/MoCo-MultiSynergy.
Yunxia Gu, Xindi Huang, Lifen Shi et al.· Bioinformatics· 0 citations
Phenotypic drug discovery enables the discovery of functional relationships between molecular structures and cellular responses. However, existing multimodal representation learning methods often optimize cross-modal alignment without considering the intrinsic organization of chemical space, resulting in distorted molecular representations and loss of structural information. We propose \textbf{PhenMol}, a structure-preserving framework for phenotype-aware molecular representation learning. PhenMol disentangles molecular and cellular representations into shared and private components, enabling phenotype-guided alignment while preserving chemical structures through a dedicated molecular branch. This design integrates cellular phenotype information without disrupting molecular neighborhood organization. Experiments on approximately $3.04 \times 10^{4}$ molecule--cell morphology pairs demonstrate that PhenMol improves molecular property prediction across 270 bioactivity tasks, molecule--phenotype retrieval, and clinical trial outcome prediction. Moreover, ECFP4-based structural analysis shows that PhenMol better preserves molecular neighborhoods and reduces embedding distortion compared with existing multimodal alignment methods. These results highlight the importance of structure-aware constraints in multimodal molecular representation learning and provide an effective approach for integrating cellular phenotypes with chemical knowledge for drug discovery.
Xuan Lin, Jingyu Sheng, Tengfei Ma et al.· 0 citations
This study provides a large-scale empirical evaluation of multimodal feature fusion for molecular property classification by systematically integrating SMILES-based chemical language representations with fingerprint-based structural descriptors across 60 benchmark datasets.
Jing Liu, Li Xue, Yin Wang et al.· Journal of Cheminformatics· 0 citations
Computer-aided drug discovery has substantially accelerated modern pharmaceutical research, where accurate molecular property prediction plays a central role in identifying promising therapeutic candidates. Self-supervised learning (SSL), which exploits large-scale unlabeled molecular data to learn transferable representations, has recently emerged as a powerful paradigm well-aligned with the data characteristics of cheminformatics. Integrating chemical domain knowledge further enhances the ability of SSL models to capture structural, physicochemical, and functional properties of molecules. In this review, we provide a systematic overview of recent advances in SSL-based molecular property prediction. We summarize representative methodological developments and analyze how multimodal molecular representation learning─by integrating sequence, graph, three-dimensional structure, and textual information─can improve the quality and expressiveness of molecular representations. We further examine the synergistic relationship between multimodal modeling and SSL, highlighting how complementary modalities can enhance representation learning in low-label settings. To demonstrate the practical benefits of multimodal molecular properties, we compare their performance with conventional SSL models on two downstream benchmark tasks with distinct prediction objectives. Finally, we discuss key open challenges, including the scarcity of high-quality 3D molecular data, modality imbalance across data sets, and the limited interpretability of learned representations. We conclude by outlining promising research directions toward more robust, generalizable, and biologically meaningful frameworks for molecular property prediction.
Shuning Yang, Lei Deng· Journal of Chemical Informat...· 0 citations