Biology subjects

GUO, J.

Publications and source records attributed to GUO, J..

7 recordsLinked to original sources

bioRxiv2026

Relation-Aware Graph-Hypergraph Networks for Antibody Sequence-Structure Co-Design

Motivation: The design of antibody complementarity-determining regions (CDRs) is pivotal for therapeutic development, yet remains challenged by the complexity of the antibody-antigen interface. Existing methods, which largely rely on pairwise relationships, struggle to capture intrinsic multi-body interactions and global structural constraints. Results: We propose the Relation-Aware Graph-Hypergraph Network (RA-GHN), a framework that unifies graph-based pairwise modeling with biophysical hypergraphs to capture high-order cooperative effects. Its heterogeneous hypergraph uses epitope-paratope interaction patches, spatial structural clusters, and high-degree hub structures. A dual-pathway architecture integrates a Relation-Aware Equivariant Graph Network with an Equivariant Hypergraph Encoder while preserving SE(3)-equivariance. We also introduce the Structure-Sequence Decoupling Index, which balances structural reliability with sequence novelty. Experiments show that RA-GHN outperforms state-of-the-art baselines across standard benchmarks and affinity optimization tasks.

bioinformatics↗

bioRxiv2026

HighFold-Lasso: Topology-Guided Prediction of Lasso Peptide Structures Using AlphaFold3

Lasso peptides are a class of ribosomally synthesized and post-translationally modified peptides (RiPPs) characterized by a mechanically interlocked topology, in which the C-terminal tail is threaded through an N-terminal macrolactam ring formed by an isopeptide bond. This unique topology often confers high thermal stability and resistance to proteolytic degradation. However, the isopeptide linkage, threaded topology, and class-specific disulfide connectivity make lasso peptide structures difficult to reconstruct reliably using general-purpose structure prediction models. Here, we developed HighFold-Lasso, which combines custom covalent connectivity with lasso scaffold-guided adaptive low-noise initialization to enable conditional structure generation guided by lasso conformational priors. On a test set of 52 experimentally determined structures, HighFold-Lasso achieved a 100% lasso topology success rate, with a mean RMSD C-alpha of 1.942 Angstrom for the ring-loop region and 2.375 Angstrom for the full peptide. The framework supports modeling of the four classical lasso peptide classes (I-IV) and can represent noncanonical amino acids, providing computational support for structure-function studies and molecular engineering of lasso peptides.

bioinformatics↗

bioRxiv2026

MINT: Predicting Enzyme-Cofactor Specificity by Multimodal Knowledge Transfer from Protein-Ligand Interactions

Protein-cofactor selectivity prediction is important for enzyme annotation, metabolic engineering, and biocatalyst discovery, but it remains challenging because experimentally supported associations are limited and unevenly distributed, and because biologically relevant differences can be subtler than those captured by general protein-ligand interaction models. Here, we present MINT (Multimodal Interaction Network with Teacher supervision), a multimodal framework for cofactor specificity prediction. MINT combines protein sequence features with cofactor sequence and graph representations and is trained with a gradual adaptation strategy that transfers useful information from a pretrained protein-ligand interaction model to the cofactor prediction task. On a fixed balanced held-out test set, MINT achieved the best overall performance among all compared methods (AUC = 0.932), and this ranking was reproduced in an independent family-stratified evaluation at the natural positive rate (pooled AUC = 0.939). Further analyses suggest that the pretrained model provides transferable structural information, while the gradual training scheme helps adapt that information to cofactor specificity. In a case study of the chemically near-identical NAD/NADP pair, MINT recovered 95.0-98.6% of true binders and resolved a single-phosphate-level distinction with direct relevance to engineering enzyme cofactor specificity. These results show that gradual transfer learning is an effective approach to cofactor specificity prediction when annotation is limited.

bioengineering↗

bioRxiv2026

NFATC2 in pancreatic cancer-associated fibroblasts predicts treatment response and facilitates ERBB-targeted therapies

Pancreatic ductal adenocarcinoma (PDAC) remains a lethal malignancy, with therapeutic resistance influenced by a dense desmoplastic stroma dominated by cancer-associated fibroblasts (CAF). Using single-cell RNA-sequencing and gene regulatory network modeling of 42 PDAC tumors, we identified a CAF subpopulation characterized by elevated NFATC2 expression that is enriched in patients with improved therapeutic response and survival. NFATC2+ CAFs exhibited tumor-suppressive features, including enhanced apoptotic signaling and suppression of ERBB pathway activity. Co-culture experiments demonstrated that NFATC2+ CAFs restrain pancreatic cancer cell growth and enhance chemotherapy-induced apoptosis, increasing sensitivity to standard-of-care chemotherapy regimens and synergizing with ERBB-targeted therapies. The favorable effect of NFATC2+ CAFs on chemotherapy response was validated in two other PDAC cohorts and in rectal cancer. Together, these findings identify NFATC2+ CAFs as a therapy-conditioned stromal state linked to improved treatment response and uncover a context-dependent vulnerability within the tumor microenvironment that may be exploited to rationally optimize combination therapies.

cancer biology↗

bioRxiv2024

Stabilized marker gene identification and functional annotation from single-cell transcriptomic data

With the rapid emergence of single-cell transcriptomics datasets, reproducible marker genes and functional annotation of cell type or state is becoming increasingly important. Conventional methods that rely on differential gene expression (DEG) analysis lack both consistency across datasets and functional annotations of selected markers. Here, we present scSCOPE, an R- based platform that utilizes stabilized LASSO (Least Absolute Shrinkage and Selection Operator) feature selection, bootstrapped co-expression networks, and pathways enrichment to identify reproducible and functionally relevant marker genes and associated pathways for cell type identification and functional annotation in scRNAseq datasets. Using 8 scRNAseq datasets of immune cell types from human and mouse tissues generated by different sequencing technologies, we show that scSCOPE outperforms other popular methods by automatically identifying marker genes and pathways with the highest consistency across all datasets. We also demonstrate that scSCOPEs gene co-expression and pathway analyses provide in-depth molecular insights into the functionality of identified marker genes. We anticipate that scSCOPE will greatly improve cell type/state annotation and accelerate the design of experimental validation and functional investigations on cell heterogeneity.

bioinformatics↗

bioRxiv2023

The highly abundant mRNA m1A modification: a new layer of gene regulation in dinoflagellates

The N1-methyladenosine (m1A) is a positively charged RNA modification known to disrupt base pairing and influence RNA stability. Despite its limited presence in the mRNA of various organism models, including yeast, mouse, and human, the exact processes of m1A biosynthesis, distribution, regulation, and function remain controversial. Dinoflagellates are a major group of single-celled eukaryotic phytoplankton having peculiar crystalline chromosomes. Their genes are arranged in unidirectional gene clusters along the chromosomes and only have minimal transcriptional regulation, implying the involvement of other critical regulatory mechanisms in gene expression. Here, we found that m1A rather than m6A is the most prevalent mRNA modification in dinoflagellates and asymmetrically distributed along mature transcripts. Utilizing the dinoflagellate species Amphidinium carterae as a study model, we identified 13481 m1A peaks characterized by a non-tRNA T-loop-like sequence motif within the transcripts of 10794 genes, many of which are involved in carbon and nitrogen metabolism. With enrichment around stop codon region and 3 UTR, dinoflagellate mRNA m1A exhibits negative correlation with translation efficiency. Notably, nitrogen depletion (N-depletion) treatment led to significant global decrease of mRNA m1A amount, causing dramatic variation in translation rates with minimal changes in transcription. Additionally, our analysis uncovered distinctive methylation patterns of m1A modification that appears to post-transcriptionally modulate gene expression through regulating translation efficiency. Thus, our findings provide the first comprehensive m1A map of dinoflagellate mRNA, shedding light on its crucial role as a post-transcriptional regulatory layer to compensate the degeneration of transcriptional regulation in dinoflagellate. This study also sets the stage for further investigation into the biogenesis and functional significance of mRNA m1A in eukaryotes.

molecular biology↗

bioRxiv2023

Mechanical Resistance to Micro-Heart Tissue Contractility unveils early Structural and Functional Pathology in iPSC Models of Hypertrophic Cardiomyopathy

Hypertrophic cardiomyopathy is the most common cause of sudden death in the young. Because the disease exhibits variable penetrance, there are likely nongenetic factors that contribute to the manifestation of the disease phenotype. Clinically, hypertension is a major cause of morbidity and mortality in patients with HCM, suggesting a potential synergistic role for the sarcomeric mutations associated with HCM and mechanical stress on the heart. We developed an in vitro physiological model to investigate how the afterload that the heart muscle works against during contraction acts together with HCM-linked MYBPC3 mutations to trigger a disease phenotype. Micro-heart muscle arrays (HM) were engineered from iPSC-derived cardiomyocytes bearing MYBPC3 loss-of-function mutations and challenged to contract against mechanical resistance with substrates stiffnesses ranging from the of embryonic hearts (0.4 kPa) up to the stiffness of fibrotic adult hearts (114 kPa). Whereas MYBPC3+/- iPSC-cardiomyocytes showed little signs of disease pathology in standard 2D culture, HMs that included components of afterload revealed several hallmarks of HCM, including cellular hypertrophy, impaired contractile energetics, and maladaptive calcium handling. Remarkably, we discovered changes in troponin C and T localization in the MYBPC3+/- HM that were entirely absent in 2D culture. Pharmacologic studies suggested that excessive Ca2+ intake through membrane-embedded channels, rather than sarcoplasmic reticulum Ca2+ ATPase (SERCA) dysfunction or Ca2+ buffering at myofilaments underlie the observed electrophysiological abnormalities. These results illustrate the power of physiologically relevant engineered tissue models to study inherited disease mechanisms with iPSC technology.

bioengineering↗