arXiv subjects

Weixiong Zhang

Publications and source records attributed to Weixiong Zhang.

At least 19 recordsLinked to original sources

arXiv2026

SPR: Toward a Graph Foundation Model for Transferable Graph Cognition via Spectral Patterns and Relational Geometry

Recently, Graph Foundation Models (GFMs) have attracted increasing attention for their potential to learn unified and generalizable knowledge across diverse graphs, thereby supporting a wide range of graph scenarios. However, unlike natural language and images, graphs lack an intuitive and unified form for perceiving and organizing transferable knowledge. Therefore, despite many initial explorations, a key problem remains unresolved: the transferable cognitive mechanism for graphs. Existing GFMs usually rely on intuitively defined mechanisms to encode transferable graph patterns, such as handcrafted structural templates (e.g., cycles and trees), fixed propagation mechanisms, and predefined random-walk patterns. These mechanisms are often prescriptive and rigid, limiting their ability to flexibly characterize diverse graph patterns across domains. This motivates the exploration of a more flexible, graph-native, and transferable cognitive mechanism. To this end, we analyze graphs from the spectral perspective and propose SPR. SPR decomposes graph information into Chebyshev polynomial bases and learns shared spectral responses, enabling unified yet adaptive cognition of continuously varying spectral patterns across graphs. We further model cross-graph relational geometry to distill recurring pairwise organization from multiple graphs, providing a shared relational reference for transferring such graph cognition across domains. Extensive experiments across diverse graphs and downstream scenarios demonstrate that SPR enables effective and transferable graph cognition.

cs.LG↗

arXiv2026

CHoE: Cross-Domain Heterogeneous Graph Prompt Learning via Structure-Conditioned Experts

Heterogeneous Graph Prompt Learning (HGPL)has emerged as a promising paradigm for bridging the gap between the objectives of pre-training foundation models and their downstream applications in heterogeneous graph settings. However, existing HGPL methods are primarily designed for in-domain scenarios, whereas real-world deployments often span multiple domains, and the data used for pre-training and downstream tasks may originate from different distributions. Consequently, the applicability of current HGPL approaches is limited to in-domain settings, and their performance typically degrades when application domains shift. To address this serious limitation, we develop CHoE, a cross-domain HGPL method built upon an expert network. During pre-training, we introduce and train structure-conditioned experts, and during prompt tuning, we adopt a structure-aware expert routing and load balancing mechanism to select structurally compatible experts for each meta-path view. In addition, we design a prompt-based semantic fusion module to integrate representations across multiple views for downstream prediction. Extensive experiments show that CHoE consistently improves performance in few-shot cross-domain applications, outperforming all baseline approaches.

cs.LG↗

arXiv2026

Unified Multi-Domain Graph Pre-training for Homogeneous and Heterogeneous Graphs via Domain-Specific Expert Encoding

Graph pre-training has achieved remarkable success in recent years, delivering transferable representations for downstream adaptation. However, most existing methods are designed for either homogeneous or heterogeneous graphs, thereby hindering unified graph modeling across diverse graph types. This separation contradicts real-world applications, where mixed homogeneous and heterogeneous graphs are ubiquitous, and distribution shifts between upstream pre-training and downstream deployment are common. In this paper, we empirically demonstrate that a balanced mixture of homogeneous and heterogeneous graph pre-training benefits downstream tasks and propose a unified multi-domain \textbf{G}raph \textbf{P}re-training method across \textbf{H}omogeneous and \textbf{H}eterogeneous graphs ($\mathbf{GPH^{2}}$). To address the lack of a unified encoder for homogeneous and heterogeneous graphs, we propose a Unified Multi-View Graph Construction that simultaneously encodes both without explicit graph-type-specific designs. To cope with the increased cross-domain distribution discrepancies arising from mixed graphs, we introduce domain-specific expert encoding. Each expert is independently pre-trained on a single graph to capture domain-specific knowledge, thereby shielding the pre-training encoder from the adverse effects of cross-domain discrepancies. For downstream tasks, we further design a Task-oriented Expert Fusion Strategy that adaptively integrates multiple experts based on their discriminative strengths. Extensive experiments on mixed graphs demonstrate that $\text{GPH}^{2}$ enables stable transfer across graph types and domains, significantly outperforming existing graph pre-training methods.

cs.LG↗

arXiv2026

Experimental Powder X-ray Diffraction Crystal Structure Determination with RealPXRD-Solver

Determining crystal structures from experimental powder X-ray diffraction data remains challenging because peak overlap, preferred orientation, and impurity phases obscure atomic arrangements. We present RealPXRD-Solver, a generative model trained on 6,250,238 theoretical structures with experiment-mimicking augmentations and a universal encoder of d-spacing--intensity fingerprints, enabling both lattice-conditioned and lattice-free inference. RealPXRD-Solver reaches a 98.3% Top-20 match rate on a 10,000-structure theoretical benchmark and achieves Top-1/Top-20 accuracies of 77.9%/91.9% on CNRS and 78.8%/92.9% on RRUFF experimental datasets, and it solved 39 previously unreported Powder Diffraction File entries.

cond-mat.mtrl-sci↗

arXiv2026

MUG: Meta-path-aware Universal Heterogeneous Graph Pre-Training

Universal graph pre-training has emerged as a key paradigm in graph representation learning, offering a promising way to train encoders to learn transferable representations from unlabeled graphs and to effectively generalize across a wide range of downstream tasks. However, recent explorations in universal graph pre-training primarily focus on homogeneous graphs and it remains unexplored for heterogeneous graphs, which exhibit greater structural and semantic complexity. This heterogeneity makes it highly challenging to train a universal encoder for diverse heterogeneous graphs: (i) the diverse types with dataset-specific semantics hinder the construction of a unified representation space; (ii) the number and semantics of meta-paths vary across datasets, making encoding and aggregation patterns learned from one dataset difficult to apply to others. To address these challenges, we propose a novel Meta-path-aware Universal heterogeneous Graph pre-training (MUG) approach. Specifically, for challenge (i), MUG introduces a input unification module that integrates information from multiple node and relation types within each heterogeneous graph into a unified representation.This representation is then projected into a shared space by a dimension-aware encoder, enabling alignment across graphs with diverse schemas.Furthermore, for challenge (ii), MUG trains a shared encoder to capture consistent structural patterns across diverse meta-path views rather than relying on dataset-specific aggregation strategies, while a global objective encourages discriminability and reduces dataset-specific biases. Extensive experiments demonstrate the effectiveness of MUG on some real datasets.

cs.LG↗

arXiv2026

LEDA: Latent Semantic Distribution Alignment for Multi-domain Graph Pre-training

Recent advances in generic large models, such as GPT and DeepSeek, have motivated the introduction of universality to graph pre-training, aiming to learn rich and generalizable knowledge across diverse domains using graph representations to improve performance in various downstream applications. However, most existing methods face challenges in learning effective knowledge from generic graphs, primarily due to simplistic data alignment and limited training guidance. The issue of simplistic data alignment arises from the use of a straightforward unification for highly diverse graph data, which fails to align semantics and misleads pre-training models. The problem with limited training guidance lies in the arbitrary application of in-domain pre-training paradigms to cross-domain scenarios. While it is effective in enhancing discriminative representation in one data space, it struggles to capture effective knowledge from many graphs. To address these challenges, we propose a novel Latent sEmantic Distribution Alignment (LEDA) model for universal graph pre-training. Specifically, we first introduce a dimension projection unit to adaptively align diverse domain features into a shared semantic space with minimal information loss. Furthermore, we design a variational semantic inference module to obtain the shared latent distribution. The distribution is then adopted to guide the domain projection, aligning it with shared semantics across domains and ensuring cross-domain semantic learning. LEDA exhibits strong performance across a broad range of graphs and downstream tasks. Remarkably, in few-shot cross-domain settings, it significantly outperforms in-domain baselines and advanced universal pre-training models.

cs.LG↗

arXiv2025

Refined HLA Linkage Disequilibrium Architectures of World Populations by a Novel Allelic Correlation Measure

Numerous diseases, particularly autoimmune disorders, are associated with the human leukocyte antigen (HLA), a small genomic region located on human chromosome 6. Adequate characterization of linkage disequilibrium (LD) in the HLA across populations is crucial for identifying genetic markers associated with specific traits and phenotypes. However, current LD measures often fail to capture HLA's structural complexity due to methodological limitations and sensitivity to low-frequency variants, marginal allele frequencies, and haplotype composition. To address these challenges, we introduced the Conditional Informatics Correlation Coefficient (CICC), which integrates conditional probability, information content, and haplotype-aware XOR logic to quantify LD robustly. When applied to high-resolution haploid genomes from the Human Pangenome Reference Consortium (HPRC), CICC revealed 10 novel high-LD regions in HLA. Further analyses using the 1000 Genomes Project and Genome Asia datasets identified nine strongly linked regions shared across five global populations-five in Class I and four in Class II. These results demonstrate CICC's ability to capture complex HLA LD structures across populations, highlighting its broad potential for disease gene mapping, population genomics, and guiding precision medicine.

q-bio.GN↗

arXiv2025

Adaptive control of dynamic networks

Real-world network systems are inherently dynamic, with network topologies undergoing continuous changes over time. Previous works often focus on static networks or rely on complete prior knowledge of evolving topologies, whereas real-world networks typically undergo stochastic structural changes that are difficult to predict in advance. To address this challenge, we define the adaptive control problem and propose an adaptive control algorithm to reduce the extra control cost caused by driver node switching. We introduce a node-level adaptive control metric to capture both the stability and consistency of each node across historical topologies. By integrating this metric with a partial matching repair strategy, our algorithm adjusts the minimum driver node set in real time at each snapshot, while minimizing unnecessary reconfigurations between consecutive time steps. Extensive experiments on synthetic and real-world dynamic networks demonstrate that the proposed adaptive control algorithm significantly outperforms the existing algorithm, reducing the switching cost by an average of 22% in synthetic networks and 19\% in real-world networks, without requiring foreknowledge of the future evolution of the network. These findings extend the theoretical scope of dynamic network controllability and open new avenues for practical applications in transportation, social, and molecular regulatory systems.

eess.SY↗

arXiv2025

From Cancer Drivers to Cancer Keepers: Paradigm Shift and Clinical Implications

Cancer research has traditionally focused on identifying driver genes, those with mutations that initiate tumorigenesis. The Cancer Driver Gene (CDG) paradigm, further supported by the observation of oncogene addiction in tumors, has successfully guided the development of targeted therapies. However, the limitations of this driver-centric view, highlighted by the broad emergence of frequent therapeutic resistance, the presence of driver mutations in healthy tissues or individuals, and the lack of identifiable drivers in many tumors, call for a shift in perspective and clinical practice. The latest network controllability perspective on cancer cells introduced the concept of Cancer Keeper Genes (CKGs) and a CKG-based paradigm for cancer therapeutics. The new concept encompasses the concept of non-oncogene addiction, emphasizing reliance on non-mutated pathways crucial for maintaining oncogenic cellular states. Here, we explore the transition towards a system-level understanding of cancer based on the CKG paradigm, emphasizing the essential role of genes required for tumor maintenance, irrespective of their initiating function or mutational capacity. We discuss clinical implications of this paradigm shift, highlighting the progress made so far and potential of targeting non-driver CKGs, genes involved in processes like DNA damage response, proteostasis, and metabolism, as a promising strategy to overcome therapeutic challenges and achieve more durable cancer control. Targeting these maintenance vulnerabilities represents a critical evolution in precision oncology, moving towards therapies designed to dismantle the networks sustaining malignancies.

q-bio.MN↗

arXiv2024

Transformer-based toxin-protein interaction analysis prioritizes airborne particulate matter components with potential adverse health effects

Air pollution, particularly airborne particulate matter (PM), poses a significant threat to public health globally. It is crucial to comprehend the association between PM-associated toxic components and their cellular targets in humans to understand the mechanisms by which air pollution impacts health and to establish causal relationships between air pollution and public health consequences. Although many studies have explored the impact of PM on human health, the understanding of the association between toxins and the associated targets remain limited. Leveraging cutting-edge deep learning technologies, we developed tipFormer (toxin-protein interaction prediction based on transformer), a novel deep-learning tool for identifying toxic components capable of penetrating human cells and instigating pathogenic biological activities and signaling cascades. Experimental results show that tipFormer effectively captures interactions between proteins and toxic components. It incorporates dual pre-trained language models to encode protein sequences and chemicals. It employs a convolutional encoder to assimilate the sequential attributes of proteins and chemicals. It then introduces a learning module with a cross-attention mechanism to decode and elucidate the multifaceted interactions pivotal for the hotspots binding proteins and chemicals. Experimental results show that tipFormer effectively captures interactions between proteins and toxic components. This approach offers significant value to air quality and toxicology researchers by allowing high-throughput identification and prioritization of hazards. It supports more targeted laboratory studies and field measurements, ultimately enhancing our understanding of how air pollution impacts human health.

cs.LG↗

arXiv2024

dnaGrinder: a lightweight and high-capacity genomic foundation model

The task of understanding and interpreting the complex information encoded within genomic sequences remains a grand challenge in biological research and clinical applications. In this context, recent advancements in large language model research have led to the development of both encoder-only and decoder-only foundation models designed to decode intricate information in DNA sequences. However, several issues persist, particularly regarding the efficient management of long-range dependencies inherent in genomic sequences, the effective representation of nucleotide variations, and the considerable computational costs associated with large model architectures and extensive pretraining datasets. Current genomic foundation models often face a critical tradeoff: smaller models with mediocre performance versus large models with improved performance. To address these challenges, we introduce dnaGrinder, a unique and efficient genomic foundation model. dnaGrinder excels at managing long-range dependencies within genomic sequences while minimizing computational costs without compromising performance. It achieves results that are not just comparable but often superior to leading DNA models such as Nucleotide Transformer and DNABERT-2. Furthermore, dnaGrinder is designed for easy fine-tuning on workstation-grade GPUs, accommodating input lengths exceeding 17,000 tokens. On a single high-performance GPU, it supports sequences longer than 140,000 tokens, making it a highly efficient and accessible tool for both basic biological research and clinical applications.

q-bio.GN↗

arXiv2023

Heterogeneous Graph Neural Networks using Self-supervised Reciprocally Contrastive Learning

Heterogeneous graph neural network (HGNN) is a very popular technique for the modeling and analysis of heterogeneous graphs. Most existing HGNN-based approaches are supervised or semi-supervised learning methods requiring graphs to be annotated, which is costly and time-consuming. Self-supervised contrastive learning has been proposed to address the problem of requiring annotated data by mining intrinsic information hidden within the given data. However, the existing contrastive learning methods are inadequate for heterogeneous graphs because they construct contrastive views only based on data perturbation or pre-defined structural properties (e.g., meta-path) in graph data while ignore the noises that may exist in both node attributes and graph topologies. We develop for the first time a novel and robust heterogeneous graph contrastive learning approach, namely HGCL, which introduces two views on respective guidance of node attributes and graph topologies and integrates and enhances them by reciprocally contrastive mechanism to better model heterogeneous graphs. In this new approach, we adopt distinct but most suitable attribute and topology fusion mechanisms in the two views, which are conducive to mining relevant information in attributes and topologies separately. We further use both attribute similarity and topological correlation to construct high-quality contrastive samples. Extensive experiments on three large real-world heterogeneous graphs demonstrate the superiority and robustness of HGCL over state-of-the-art methods.

cs.LG↗

arXiv2023

Total controllability analysis discovers explainable drugs for Covid-19 treatment

Network medicine has been pursued for Covid-19 drug repurposing. One such approach adopts structural controllability, a theory for controlling a network (the cell). Motivated to protect the cell from viral infections, we extended this theory to total controllability and introduced a new concept of control hubs. Perturbation to any control hub renders the cell uncontrollable by exogenous stimuli, e.g., viral infections, so control hubs are ideal drug targets. We developed an efficient algorithm for finding all control hubs and applied it to the largest homogenous human protein-protein interaction network. Our new method outperforms several popular gene-selection methods, including that based on structural controllability. The final 65 druggable control hubs are enriched with functions of cell proliferation, regulation of apoptosis, and responses to cellular stress and nutrient levels, revealing critical pathways induced by SARS-CoV-2. These druggable control hubs led to drugs in 4 major categories: antiviral and anti-inflammatory agents, drugs on central nerve systems, and dietary supplements and hormones that boost immunity. Their functions also provided deep insights into the therapeutic mechanisms of the drugs for Covid-19 therapy, making the new approach an explainable drug repurposing method. A remarkable example is Fostamatinib that has been shown to lower mortality, shorten the length of ICU stay, and reduce disease severity of hospitalized Covid-19 patients. The drug targets 10 control hubs, 9 of which are kinases that play key roles in cell differentiation and programmed death. One such kinase is RIPK1 that directly interacts with viral protein nsp12, the RdRp of the virus. The study produced many control hubs that were not targets of existing drugs but were enriched with proteins on membranes and the NF-$κ$B pathway, so are excellent candidate targets for new drugs.

q-bio.MN↗

arXiv2023

NetMoST: A network-based machine learning approach for subtyping schizophrenia using polygenic SNP allele biomarkers

Subtyping neuropsychiatric disorders like schizophrenia is essential for improving the diagnosis and treatment of complex diseases. Subtyping schizophrenia is challenging because it is polygenic and genetically heterogeneous, rendering the standard symptom-based diagnosis often unreliable and unrepeatable. We developed a novel network-based machine-learning approach, netMoST, to subtyping psychiatric disorders. NetMoST identifies polygenic risk SNP-allele modules from genome-wide genotyping data as polygenic haplotype biomarkers (PHBs) for disease subtyping. We applied netMoST to subtype a cohort of schizophrenia subjects into three distinct biotypes with differentiable genetic, neuroimaging and functional characteristics. The PHBs of the first biotype (36.9% of all patients) were related to neurodevelopment and cognition, the PHBs of the second biotype (28.4%) were enriched for neuroimmune functions, and the PHBs of the third biotype (34.7%) were associated with the transport of calcium ions and neurotransmitters. Neuroimaging patterns provided additional support to the new biotypes, with unique regional homogeneity (ReHo) patterns observed in the brains of each biotype compared with healthy controls. Our findings demonstrated netMoST's capability for uncovering novel biotypes of complex diseases such as schizophrenia. The results also showed the power of exploring polygenic allelic patterns that transcend the conventional GWAS approaches.

q-bio.MN↗

arXiv2022

Pixel-global Self-supervised Learning with Uncertainty-aware Context Stabilizer

We developed a novel SSL approach to capture global consistency and pixel-level local consistencies between differently augmented views of the same images to accommodate downstream discriminative and dense predictive tasks. We adopted the teacher-student architecture used in previous contrastive SSL methods. In our method, the global consistency is enforced by aggregating the compressed representations of augmented views of the same image. The pixel-level consistency is enforced by pursuing similar representations for the same pixel in differently augmented views. Importantly, we introduced an uncertainty-aware context stabilizer to adaptively preserve the context gap created by the two views from different augmentations. Moreover, we used Monte Carlo dropout in the stabilizer to measure uncertainty and adaptively balance the discrepancy between the representations of the same pixels in different views.

cs.CV↗

arXiv2022

RAW-GNN: RAndom Walk Aggregation based Graph Neural Network

Graph-Convolution-based methods have been successfully applied to representation learning on homophily graphs where nodes with the same label or similar attributes tend to connect with one another. Due to the homophily assumption of Graph Convolutional Networks (GCNs) that these methods use, they are not suitable for heterophily graphs where nodes with different labels or dissimilar attributes tend to be adjacent. Several methods have attempted to address this heterophily problem, but they do not change the fundamental aggregation mechanism of GCNs because they rely on summation operators to aggregate information from neighboring nodes, which is implicitly subject to the homophily assumption. Here, we introduce a novel aggregation mechanism and develop a RAndom Walk Aggregation-based Graph Neural Network (called RAW-GNN) method. The proposed approach integrates the random walk strategy with graph neural networks. The new method utilizes breadth-first random walk search to capture homophily information and depth-first search to collect heterophily information. It replaces the conventional neighborhoods with path-based neighborhoods and introduces a new path-based aggregator based on Recurrent Neural Networks. These designs make RAW-GNN suitable for both homophily and heterophily graphs. Extensive experimental results showed that the new method achieved state-of-the-art performance on a variety of homophily and heterophily graphs.

cs.LG↗

arXiv2022

Identification of cancer-keeping genes as therapeutic targets by finding network control hubs

Finding cancer driver genes has been a focal theme of cancer research and clinical studies. One of the recent approaches is based on network structural controllability that focuses on finding a control scheme and driver genes that can steer the cell from an arbitrary state to a designated state. While theoretically sound, this approach is impractical for many reasons, e.g., the control scheme is often not unique and half of the nodes may be driver genes for the cell. We developed a novel approach that transcends structural controllability. Instead of considering driver genes for one control scheme, we considered control hub genes that reside in the middle of a control path of every control scheme. Control hubs are the most vulnerable spots for controlling the cell and exogenous stimuli on them may render the cell uncontrollable. We adopted control hubs as cancer-keep genes (CKGs) and applied them to a gene regulatory network of bladder cancer (BLCA). All the genes on the cell cycle and p53 singling pathways in BLCA are CKGs, confirming the importance of these genes and the two pathways in cancer. A smaller set of 35 sensitive CKGs (sCKGs) for BLCA was identified by removing network links. Six sCKGs (RPS6KA3, FGFR3, N-cadherin (CDH2), EP300, caspase-1, and FN1) were subjected to small-interferencing-RNA knockdown in four cell lines to validate their effects on the proliferation or migration of cancer cells. Knocking down RPS6KA3 in a mouse model of BLCA significantly inhibited the growth of tumor xenografts in the mouse model. Combined, our results demonstrated the value of CKGs as therapeutic targets for cancer therapy and the potential of CKGs as an effective means for studying and characterizing cancer etiology.

q-bio.MN↗

arXiv2022

Response to: Significance and stability of deep learning-based identification of subtypes within major psychiatric disorders. Molecular Psychiatry (2022)

Recently, Winter and Hahn [1] commented on our work on identifying subtypes of major psychiatry disorders (MPDs) based on neurobiological features using machine learning [2]. They questioned the generalizability of our methods and the statistical significance, stability, and overfitting of the results, and proposed a pipeline for disease subtyping. We appreciate their earnest consideration of our work, however, we need to point out their misconceptions of basic machine-learning concepts and delineate some key issues involved.

cs.LG↗