This paper investigates a novel and critical problem of Graph Cross-Domain Continual Fine-Tuning, which aims to adapt a large pre-trained Graph Foundation Model across diverse domains. Existing continual graph learning methods are mostly limited to incremental settings within only a single domain, and are typically trained from scratch. As a result, they fail to handle cross-domain shifts effectively, suffer from severe forgetting, and lack transferability. To address these challenges, we present G-CORMoL, Graph Continual Fine-tuning with Orthogonal, Router-driven Mixture of LoRA experts. G-CORMoL achieves effective adaptation while preserving prior knowledge by enforcing mathematical orthogonality between expert LoRA adapters, thereby eliminating interference across tasks. It further supports cross-domain knowledge transfer through a symmetric dual-driven routing mechanism that learns a global composition policy over all learned LoRA experts. In addition, it promotes expert specialization via a contrastive objective with theoretical guarantees. Extensive experiments on different cross-domain task orders demonstrate that G-CORMoL achieves robust state-of-the-art performance, not only preventing catastrophic forgetting but also leveraging accumulated knowledge to enable positive transfer.

This paper investigates a novel and critical problem of Graph Continual Test-Time Training, which aims to enable a frozen pre-trained graph model to adapt continuously to evolving out-of-distribution (OOD) graphs without supervision. Existing test-time training methods primarily focus on one-step adaptation and overlook long-term knowledge retention, while conventional continual learning approaches rely on labeled data and static memory replay. Consequently, they are unable to handle sequential OOD domains effectively, often suffering from severe forgetting and limited efficiency in dynamic graph environments. To address these challenges, we propose DPCGL (Dynamic Prompts-based Continual Graph Learning), a data-centric framework that performs continual test-time training through adaptive prompt optimization. DPCGL freezes the pre-trained backbone and maintains a dynamic prompt pool, where prompts are adaptively selected and updated for each incoming graph domain. This design enables parameter-efficient adaptation and mitigates forgetting by organizing transferable knowledge within prompts rather than model weights. Furthermore, DPCGL jointly optimizes three objectives: similarity alignment for representativeness, KL divergence regularization for knowledge preservation, and diversity constraint for generalization, providing both stability and adaptability during continual adaptation. Extensive experiments on multiple evolving OOD graph benchmarks demonstrate that DPCGL achieves state-of-the-art performance, effectively alleviating catastrophic forgetting and enabling robust continual adaptation across domains.

This paper addresses the challenge of graph domain adaptation on evolving, multiple out-of-distribution (OOD) graphs. Conventional graph domain adaptation methods are confined to single-step adaptation, making them ineffective in handling continuous domain shifts and prone to catastrophic forgetting. This paper introduces the Graph Continual Adaptive Learning (GCAL) method, designed to enhance model sustainability and adaptability across various graph domains. GCAL employs a bilevel optimization strategy. The "adapt" phase uses an information maximization approach to fine-tune the model with new graph domains while re-adapting past memories to mitigate forgetting. Concurrently, the "generate memory" phase, guided by a theoretical lower bound derived from information bottleneck theory, involves a variational memory graph generation module to condense original graphs into memories. Extensive experimental evaluations demonstrate that GCAL substantially outperforms existing methods in terms of adaptability and knowledge retention.

Deep learning techniques have driven significant progress in various analytical tasks within 3D genomics in computational biology. However, a holistic understanding of 3D genomics knowledge remains underexplored. Here, we propose ***MIX-HIC***, the first multimodal foundation model of 3D genome that integrates both 3D genome structure and epigenomic tracks, which obtains unified and comprehensive semantics. For accurate heterogeneous semantic fusion, we design the cross-modal interaction and mapping blocks for robust unified representation, yielding the accurate aggregation of 3D genome knowledge. Besides, we introduce the first large-scale dataset comprising over ***1 million*** pairwise samples of Hi-C contact maps and epigenomic tracks for high-quality pre-training, enabling the exploration of functional implications in 3D genomics. Extensive experiments show that MIX-HIC significantly surpasses existing state-of-the-art methods in diverse downstream tasks. This work provides a valuable resource for advancing 3D genomics research.