Proteolysis Targeting Chimeras (PROTACs) are heterobifunctional ligands bridging Proteins‐Of‐Interest (POIs) and E3 ligases for ubiquitin‐proteasome degradation, promising to target the ‘undruggable’. While PROTAC research primarily relies on costly and time‐consuming wet‐lab experiments, deep learning offers potential to accelerate development and reduce expenses. However, many deep learning methods for PROTAC degradation prediction overlook hierarchical molecular representation and protein structural data, hindering data modeling. Moreover, their black‐box nature hampers interpretability, failing to provide intuitive insights into substructure interactions within the PROTAC system. This study introduces PROTAC‐STAN, a structure‐informed deep ternary attention network (STAN) framework for interpretable PROTAC degradation prediction. PROTAC‐STAN represents PROTAC molecules across atom, molecule, and property hierarchies and incorporates structure information for POIs and E3 ligases via protein language model. Furthermore, it simulates interactions among three entities at substructure levels via a novel ternary attention network tailored for the PROTAC system, providing unprecedented insights into the degradation mechanism. PROTAC‐STAN yields over 10% improvement across multiple metrics compared to the best baselines, while providing significant interpretability through atomic and residue level visualization of molecule and complex. Exploratory evaluations and case studies demonstrate strong real‐world applicability. The excellence of PROTAC‐STAN is anticipated to establish a foundational tool for future PROTAC research.

Single‐cell RNA sequencing (scRNA‐seq) has unveiled extensive cellular heterogeneity, yet precise cell type annotation and the identification of novel cell populations remain significant challenges. scHeteroNet, a novel graph neural network framework specifically designed to leverage heterophily in scRNA‐seq data, is presented. Unlike traditional methods that assume homophily, scHeteroNet captures complex cell–cell interactions by integrating information from both immediate and extended cellular neighborhoods, resulting in highly accurate cell representations. Additionally, scHeteroNet incorporates an innovative novelty propagation mechanism that robustly detects previously uncharacterized cell types. Comprehensive evaluations across diverse scRNA‐seq datasets demonstrate that scHeteroNet consistently outperforms state‐of‐the‐art approaches in both cell type classification and novel cell detection. This heterophily‐aware approach enhances the ability to uncover cellular diversity, providing deeper insights into complex biological systems and advancing the field of single‐cell analysis.

Post-translational modifications (PTMs) form a complex combinatorial “code” that orchestrates protein function and cellular signaling. However, deciphering this code by predicting PTM sites and linking sites to their regulatory enzymes remains a fundamental challenge. Here, we present COMPASS-PTM, a mechanism-aware, coarse-to-fine learning framework that unifies residue-level multi-label PTM prediction with enzyme-substrate assignment by jointly modeling PTM patterns and their catalytic regulators. COMPASS-PTM builds upon protein language models, integrating physicochemical descriptors and a crosstalk-aware prompting mechanism to learn biologically coherent patterns of cooperative and antagonistic modifications, while addressing the dual long-tail distribution inherent in PTM data. Across multiple proteome-scale benchmarks, COMPASS-PTM improves over the strongest evaluated baselines, with a 122% relative improvement in F1-score for multi-label site prediction and a 54% gain in zero-shot enzyme assignment. Furthermore, the model demonstrates interpretable generalization, recovering canonical kinase motifs and mechanistically linking missense variants to both local PTM disruptions and global rewiring of enzyme-substrate networks. By coupling statistical learning with explicit biochemical knowledge, COMPASS-PTM unifies site-level and enzyme-level prediction into a single framework that learns the grammar underlying protein regulation and signaling.

Automated segmentation of abdominal organs and tumors in medical images is a challenging yet essential task in medical image analysis. Deep learning has shown excellent performance in many medical image segmentation tasks, but most prior efforts were fragmented, addressing individual organ and tumor segmentation tasks with specialized networks. To tackle the challenges of abdominal organ and tumor segmentation using partially labeled datasets, we introduce Re-parameterizing Mixture-of-Diverse-Experts (RepMode) to abdominal organ and tumor segmentation. Within the RepMode framework, the Mixture-of-Diverse-Experts (MoDE) block forms the foundation, learning generalized parameters applicable across all tasks. We seamlessly integrate the MoDE block into a U-shaped network with dynamic heads, addressing multi-scale challenges by dynamically combining experts with varying receptive fields for each organ and tumor. Our framework incorporates task encoding in both the encoder–decoder section and the segmentation head, enabling the network to adapt throughout the entire system based on task-related information. We evaluate our approach on the multi-organ and tumor segmentation (MOTS) dataset. Experiments show that DoDRepNet outperforms previous methods, including multi-head networks and single-network approaches, giving a highly competitive performance compared with the original single network with dynamic heads. DoDRepNet offers a promising approach to address the complexities of abdominal organ and tumor segmentation using partially labeled datasets, enhancing segmentation accuracy and robustness.