Biology subjects

Seo, S.

Publications and source records attributed to Seo, S..

At least 19 recordsLinked to original sources

bioRxiv2026

Predicting conformational changes induced by biomolecular interactions

Biomolecular interactions trigger structural changes of protein complexes essential for cellular signaling, regulation, and drug action. Modern co-folding models can now generate detailed structures of multi-component complexes directly from sequence. However, recent evaluations reveal that they frequently fail to capture specific conformational states induced by a ligand or partner. Here, we show that explicitly modeling the structural transition from an unbound to a bound state, rather than generating the bound complex alone, better resolves these interaction-induced states. Our generative learning framework, K-Fold, transforms independently predicted unbound component priors into an assembled complex through an ultrafast diffusion bridge, operating entirely without multiple sequence alignments. By eliminating co-evolutionary dependencies, K-Fold accurately recovers non-coevolved interfaces in antibody-antigen and targeted-degrader complexes, as well as ligand-directed regulatory states in GPCRs and kinases. Furthermore, by coupling this diffusion bridge with diverse monomeric equilibrium ensembles, K-Fold provides a computationally tractable foundation for sampling complex multi-state landscapes without retraining. These results demonstrate that formulating molecular assembly as an explicit structural transition moves structure prediction beyond static geometries toward capturing biomolecular dynamics.

molecular biology↗

bioRxiv2026

ELiAH2: update of the atlas of E3 ligases in human tissues by integrating tissue-specificity and docking properties of ligands for targeted protein degradation

Targeted protein degradation (TPD), including proteolysis-targeting chimeras (PROTACs), requires coordinated selection of E3 ligases, target proteins, and ligands to achieve efficient and tissue-selective degradation. However, current TPD resources provide limited support for integrating tissue-specific transcriptomic evidence with structural compatibility during early candidate prioritization. Here, we present ELiAH2, an expanded database for transcriptome- and structure-guided prioritization of TPD candidates. ELiAH2 integrates GTEx RNA-seq profiles across 24 human tissues to characterize tissue-specific expression patterns of E3 ligases and target genes. Compared with the previous version, ELiAH, current update presented over 200% of expansion for the number of covered tissues and samples. Previously, E3 ligase-target gene associations were inferred using mutual information-based network analysis, enabling users to explore transcriptomic relationships. To further support structural evaluation, ELiAH2 provides large-scale docking datasets comprising 827,385 E3 ligase-target protein pairs and 7,800,521 protein-ligand docking models, including E3 ligase-ligand and target protein-ligand interactions. Docking scores, ranked interaction tables, gene and ligand identifiers, and downloadable PDB structures are incorporated into the web interface. By combining tissue specificity, E3-target association, protein-protein docking, and protein-ligand docking, ELiAH2 provides a comprehensive resource for narrowing the TPD search space and supporting rational degrader design. ELiAH2 is freely available at https://eliahdb.org.

bioinformatics↗

bioRxiv2026

Synthesis-aware generative design in trillion-scale chemical spaces for automated drug discovery

Autonomous drug discovery via generative molecular design is critically bottlenecked by the production of chemically intractable structures. Multi-trillion-scale make-on-demand libraries guarantee synthetic feasibility, but conventional virtual screening cannot efficiently navigate these vast spaces or address multi-parameter optimization. We introduce Hyper Screening X, a generative AI framework integrating structure-based design with automated synthesis hardware. By encoding deterministic reaction logic in a generative flow network, the framework evaluates only 10 million candidates to navigate an 11-trillion-compound space, jointly optimizing physicochemical properties while ensuring compatibility with automated synthesis. Prospective validation against an SLC1A5 variant with a cryptic interface yielded a 96% synthesis success rate and a 50% functional hit rate, identifying two first-in-class lead compounds. This synthesis-aware strategy advances autonomous, closed-loop drug discovery.

bioinformatics↗

bioRxiv2026

Microfluidic Capture-Enrichment of individualized microbes from urban wastewater reveals a hidden reservoir of potential pre-emergent pathogens

Current environmental biosurveillance infrastructure excels at tracking known biological threats but is limited in its ability to proactively identify potential pre-emergent pathogens (PEPs). To address this gap, we developed a microfluidics-based Capture-Enrichment pipeline that isolates individual environmental cells via microdroplet encapsulation, preserving community biodiversity while enabling iterative selection against human sera. When applied to urban wastewater, this approach successfully recovered pathogens, commensals, and PEPs that fell below the detection limits of conventional amplicon sequencing. Genomic analysis revealed that wastewater-derived ESKAPE pathogens cluster closely with clinical isolates, showing that the pipeline captures clinically relevant threats. Furthermore, isolated PEPs exhibited known and emerging phenotypes associated with early-stage pathogenesis and antimicrobial resistance. These findings demonstrate that targeting PEPs can fundamentally expand the scope and quality of biosurveillance by providing a scalable strategy to identify, characterize, and close important gaps in our understanding of emerging microbial pathogens.

microbiology↗

bioRxiv2026

Integrated morphological, multi-omics, and functional profiling reveals microglial plasticity driven by AD risk genes

Microglia, the resident macrophages of the central nervous system, are highly dynamic cells essential for brain homeostasis. While genome-wide association studies (GWAS) strongly implicate microglial dysfunction in Alzheimer's disease (AD), the mechanistic links coordinating their diverse transcriptional, morphological, and functional states remain poorly understood. Using human induced pluripotent stem cell-derived microglia (iMicroglia) and single-cell RNA sequencing, we identified six distinct transcriptional profiles and mapped them to specific morphological phenotypes via targeted immunofluorescence, establishing a link between microglial morphology and molecular identity. Transcriptomic and morphological profiling further demonstrated profound microglial plasticity, revealing distinct, stimulus-specific responses to AD-relevant pathologies, including Tau PFF, amyloid-beta, and apoptotic neurons. To assess how AD risk variants perturb these states, we performed high-efficiency CRISPR-Cas9 ribonucleoprotein (RNP) knockouts of specific AD GWAS genes. Bulk RNA-seq profiling revealed extensive transcriptional remodeling following genetic perturbation. Crucially, we show that depletion of these AD risk genes disrupts baseline morpho-transcriptomic coupling and fundamentally alters microglial phagocytic capacity. Together, this study reveals how AD GWAS genes may drive microglia into dysfunctional states characterized by altered morphology, distinct multi-omic signatures, and impaired functions.

neuroscience↗

bioRxiv2026

AtlasFold: Protein structure prediction with metagenomic-scale language models

Protein language models (PLMs) trained on evolutionary sequences learn representations that encode protein structure, enabling direct structure prediction without multiple-sequence alignments (MSAs). Here we present the Atlas model family, an open and trainable system spanning protein language modeling, monomer folding, and protein-complex prediction. AtlasLM-3B is a 3B-scale language model trained with masked language modeling on approximately 1.56 billion sequences, including metagenomic data, and outperforms the similarly sized ESM2-3B in unsupervised contact prediction. Building on these representations, AtlasFold predicts all-atom protein structures and achieves state-of-the-art accuracy among PLM-based folding models. Fine-tuning AtlasFold for protein-complex prediction produces AtlasFold-Multimer (AtlasFold-M), whose antibody-antigen prediction performance is comparable to that of AlphaFold3 and ESMFold2. This protein-specific folding architecture enables fast, memory-efficient inference with AtlasFold and AtlasFold-M. By releasing the training code and data, stage checkpoints, and model weights under the MIT License, we provide a foundation for advancing PLM-based protein structure prediction.

bioinformatics↗

bioRxiv2026

Elucidating enzyme-substrate specificity through co-folding foundation model

Enzymatic catalysis relies on precise structural and chemical complementarity, yet systematically mapping enzyme-substrate interactions remains a critical bottleneck. While structure-aware methods have advanced functional annotation, their reliance on predefined binding pockets and rigid-body docking fails to capture the ligand-induced conformational changes essential for catalytic turnover. Here we introduce Boltz2ESI, an end-to-end framework that predicts enzyme-substrate interactions by leveraging structural knowledge learned by a biomolecular foundation model. Through native co-folding, the framework inherently captures active-site plasticity without requiring predefined pocket annotations. Integrating these learned biophysical priors with global evolutionary context and geometric molecular descriptors, Boltz2ESI consistently outperforms state-of-the-art sequence-based and rigid-docking approaches. Extensive validation demonstrates that the framework accurately discriminates tight sub-family specificities, enabling effective candidate prioritization for biosynthetic pathway elucidation, as demonstrated on the withanolide pathway. Ultimately, this structure-dynamic approach establishes an actionable foundation for accelerating rational biocatalyst discovery and large-scale pathway de-orphaning.

bioinformatics↗

bioRxiv2026

Patient-specific zebrafish models reveal a complex phenotypic spectrum in NBAS-associated Atypical Osteogenesis Imperfecta

Neuroblastoma amplified sequence gene (NBAS) variants are associated with short stature, optic atrophy, and Pelger-Huet anomaly (SOPH) syndrome. We previously identified compound heterozygous variants in NBAS to cause atypical Osteogenesis Imperfecta (OI), with these patients presenting with short stature, developmental delay and recurrent long-bone fractures. However, skeletal disease progression due to these variants and the disease mechanisms underlying NBAS-associated OI remain poorly understood. Here, we provide a clinical update on previously identified patients and investigate the role of NBAS during skeletal development using zebrafish knockout and patient-specific missense variant zebrafish models. Homozygous knockout larvae exhibited delayed operculum development, reduced bone ossification, and defects in Meckels cartilage morphology and its underlying cellular structure. Homozygous missense larvae displayed milder cartilage defects without any major defects to early skeletal structures. Seemingly opposing phenotypes were observed in compound heterozygous zebrafish carrying the knockout and missense alleles in trans, with no obvious phenotypes seen in the Meckels cartilage and accelerated operculum development observed. Together, our results demonstrate that different nbas variants differentially affect skeletal development, suggesting complex spectrums of phenotypic and pathogenic mechanisms in NBAS-associated atypical OI.

genetics↗

bioRxiv2026

Computational Design Strategies for Nanoscale 3D Auxetic Metastructures from DNA

Auxetic metamaterials exhibit negative Poissons ratio behaviors due to their architecture of periodically arranged unit cells. Although mechanical metamaterials are well established at the macroscale, programmable auxetic units remain scarce at the nanoscale. DNA origami offers a promising platform to bridge this gap, but design principles for dynamically deformable 3D auxetic nanostructures remain largely unexplored. Here, we develop design strategies for such 3D auxetic metastructures built from wireframe DNA origami. As a model system, we use a 3D re-entrant triangular unit composed of double-stranded DNA (dsDNA) bundle edges connected by single-stranded DNA (ssDNA) joints. Using coarse-grained molecular dynamics (MD) and umbrella-sampling free-energy simulations, we examine how edge design and joint-connection scheme govern auxetic responses and the energetics of the structural transformation. Our results show that auxetic performance and deformation energetics emerge from the coupled effects of DNA bundle rigidity and connector mechanics at the joints. This study provides mechanistic insights and design guidelines for programmable auxetic motion and energetics in 3D DNA origami metamaterials, advancing the development of stimuli-responsive nanomechanical devices.

biophysics↗

bioRxiv2026

Functional surrogacy enables Vascular Ehlers-Danlos Syndrome modelling in zebrafish in the absence of a COL3A1 ortholog

Pathogenic variants in COL3A1 cause Vascular Ehlers-Danlos syndrome (vEDS), a rare connective tissue disorder characterised by vascular fragility, increasing the risk of arterial ruptures/dissection. Advances in genomic sequencing have led to an increasing number of COL3A1 variants where the clinical significance is unclear, with these being termed variants of uncertain significance (VUS). VUS creates challenges for diagnosis and clinical management. Thus major efforts have been made to reclassify these to either pathogenic or benign variants in disease causality. Functional data from model systems can provide significant evidence to clinicians on the pathogenicity of a variant. To address the increasing numbers of VUS in COL3A1, we developed a fast pipeline using F0 crispant zebrafish to provide functional evidence for variant classification despite there being no direct orthologue of COL3A1 in zebrafish. Loss of col5a1 resulted in cardiac defects, dysmorphic blood vessel structures and delayed angiogenic sprouting. Trunk haemorrhage prevalence under physical stress increased in col5a1 knockout zebrafish, recapitulating vEDS patients. Remarkably, co-injection of F0 col5a1 knockout crispants with human wildtype COL3A1 mRNA partially rescued cardiac and vascular phenotypes, indicating a level of functional conservation between zebrafish type V and human type III collagen. These findings establish a tractable in vivo platform for functional assessment of COL3A1 VUS. Phenotypic rescue with wildtype COL3A1 provides a benchmark against which the pathogenicity of variants can be evaluated, generating functional evidence for VUS reclassification. Our model provides both a valuable tool for investigating vEDS disease mechanisms and a clinically relevant platform to improve diagnoses for patients with suspected vEDS.

genetics↗

bioRxiv2026

Autosomal dominant CDC45 deficiency with allelic expression bias causes a novel genetic disease of the immune system

Here we describe a damaging heterozygous variant in CDC45 in an individual with common variable immunodeficiency (CVID) and recurrent viral infections. This individual has variably decreased number of circulating NK cells, disruption in the ratio of CD56bright to CD56dim cells, and consistently decreased NK cell function. Interestingly, the inherited CDC45 variant is also present in a sibling with less severe clinical manifestations; we determined that allelic bias of the damaging allele accounts for this differential expressivity. As previously reported for other helicase variants that cause inborn errors of immunity (IEI), we found cell cycle defects in immune cells from the proband that lead to reduced survival of NK cells. Together, these findings link another member of the core replicative helicase complex to inborn errors of immunity and highlight the sensitivity of NK cells to these variants. They also define another case of allelic bias contributing to variable expressivity of an IEI gene.

immunology↗

bioRxiv2026

Effects of Cognitive Demand Reduction on Choice Overload

Choice overload occurs when an ever-growing number of options impairs decision quality, because evaluating options taxes cognitive resources. We investigated whether reducing cognitive demand could mitigate overload by encouraging greater cognitive effort to achieve optimal choice. We conducted two experiments manipulating cognitive demand in complementary ways: Experiment 1 reduced demand by presenting high-attractiveness sets, and Experiment 2 did so by providing a shortlist tool. In both experiments, participants chose from sets of 6-24 options while their eye-gaze and electroencephalographic (EEG) data were recorded. We found that reducing demand made decisions faster, but did not improve choice performance as set-size increased. Under low-demand conditions, eye-gaze measures revealed narrower search and EEG measures showed reduced working memory engagement per option, together indicating less searching and processing efforts. These results suggest that even with reduced cognitive demand, people coast through easier decisions, conserving effort and leaving the choice overload effect largely intact.

neuroscience↗

bioRxiv2025

COBRA: Cell-type-specific Orthogonal Batch effect Removal Algorithm in single cell RNA-sequencing data

Single-cell RNA sequencing (scRNA-seq) enables high-resolution profiling of cellular heterogeneity, yet batch effects remain a critical challenge in data integration. Existing batch correction methods often assume homogeneous batch effect across cell types, operate in reduced-dimensional space leading to potential loss of biological information, and require extensive computational resources. Here, we introduce COBRA (https://github.com/wonlab-healthstat/COBRA), a linear model-based batch correction method that explicitly adjusts cell-type-specific batch effect. By orthogonalizing batch-associated parameters with respect to biological variables, COBRA removes technical artifacts while preserving biologically meaningful transcriptional differences. When cell type annotations are unavailable, COBRA implements an iterative clustering algorithm to estimate pseudo-cell types while accounting for batch effects. COBRA retains the full gene expression matrix, ensuring seamless integration for downstream analyses. We evaluated COBRA across simulated and real-world datasets, including type 2 diabetes, and COVID-19 datasets. COBRA outperformed in terms of batch mixing efficiency, preservation of biological group structure, and accuracy of differentially expressed gene detection.

bioinformatics↗

bioRxiv2025

Tripartite AAV Systems for EYS Retinal Gene Therapy

Mutations in the Eyes Shut Homolog (EYS) gene are a leading cause of autosomal recessive retinitis pigmentosa, a progressive retinal degenerative disease for which no effective treatment currently exists. However, the large size of the EYS coding sequence ([~]9.4 kb) exceeds the packaging limit of adeno-associated virus (AAV) vectors, posing a major barrier to gene replacement therapy. To address this challenge, we developed a tripartite AAV vector system that enables delivery and reconstitution of the full-length EYS gene using a Cre-lox-based unidirectional DNA recombination strategy, Uni-STAR (Uni-directional and Site-specific Transgene Assembly by Recombination). The system consists of three AAV constructs carrying discrete EYS segments flanked by engineered, non-compatible lox sites that drive ordered and unidirectional recombination in target cells. We validated this system in vitro by demonstrating successful reconstitution and expression of full-length EYS protein in HEK293T cells. In vivo, subretinal co-injection of the three AAV vectors into mouse eyes led to precise reconstitution and expression of full-length EYS protein in the retina. These findings establish the feasibility of using a tripartite AAV system to deliver the complete EYS gene and provide a foundation for future therapeutic development targeting EYS-associated retinal degenerations.

genetics↗

bioRxiv2025

A Nanoscale Jitterbug Transformer from DNA

Many viruses have intricate polyhedral shells capable of symmetric transformations in response to external stimuli to initiate payload release. Such deployable auxetic nanostructures are not available in the synthetic realm. Here we present a nanoscale Jitterbug transformer using DNA origami that can reconfigure its structure upon chemical and optical signals while maintaining a Poissons ratio of -1. By leveraging molecular dynamics simulations, we design the Jitterbug DNA to form a compact octahedron by storing elastic energy and spontaneously transition into an expanded cuboctahedron by releasing it. DNA transformers are explored like viruses that can create nanopores on lipid membranes and regulate payload release into vesicles. This work integrates programmable DNA self-assembly with free-energy-guided mechanical design, providing a pathway toward adaptive nanomaterials with potential in synthetic organelles and stimuli-responsive nanodevices.

biophysics↗

bioRxiv2025

Understanding Structural Mechanics of Ligated DNA Crystals via Molecular Dynamics Simulation

DNA self-assembly is a highly programmable method that can construct arbitrary architectures based on sequence complementarity. Among various constructs, DNA crystals are macroscopic crystalline materials formed by assembling motifs via sticky end association. Due to their high structural integrity and size ranging from tens to hundreds of micrometers, DNA crystals offer unique opportunities to study structural properties and deformation behaviors of DNA assemblies. For example, enzymatic ligation of sticky ends can selectively seal nicks resulting in more robust structures with enhanced mechanical properties. However, the research efforts have been mostly on experiments such as different motif designs, structural optimization, or new synthesis methods, while their mechanics are not fully understood. The complex properties of DNA crystals are difficult to study via experiment alone, and numerical simulation can complement and inform the experiment. Coarse-grained molecular dynamics (MD) simulation is a powerful tool that can probe the mechanics of DNA assemblies. Here, we investigate DNA crystals made of four different motif lengths with various ligation patterns (full ligation, major directions, connectors, and in-plane) using oxDNA, an open-source, coarse-grained MD platform. We find that several distinct deformation stages emerge in response to mechanical loading and that the number and the location of the ligated nucleotides can significantly modulate structural behaviors. These findings should be useful for predicting crystal properties and thus improving the design.

biophysics↗

bioRxiv2025

Blue Light Sonata: Dynamic variation of red:blue ratio during the photoperiod differentially affects leaf photosynthesis, pigments, and growth in lettuce

Vertical farming (VF) has unparalleled capacity to highly customize plant growth environments. In VF, red and blue LED lights are predominantly used as the two main wavelengths for photosynthesis. For many plants, red light increases biomass, and blue light can increase nutritional content. Because red light is more cost-and energy-efficient to produce than blue light, refined growth recipes are imperative to mutualistically improve efficiency with crop yield and quality. This studys aim was to balance lighting energy-use with growth and nutritional quality by using "dynamic lighting" recipes to reduce durations of high-intensity blue light. Lettuce (Lactuca sativa L.) was grown for 21 days at 220 mol m-2 s-1, receiving one of five R:B ratios (R:B100:0, R:B95:5, R:B89:11, R:B50:50, and R:B0:100) for either the whole 18-hour photoperiod (Whole Day), the first six hours of the photoperiod (Morning), or the last six hours of the photoperiod (Evening). Morning and Evening treatments received low blue (R:B89:11) for the remaining 12 hours of the day. The Morning and Evening high blue treatments had greater fresh weight and leaf area than their respective Whole Day treatments, attributed to reduced instantaneous leaf photosynthesis under high blue. High blue reduced photosynthesis during only the six hours of Morning and Evening treatments, compared to the full impact of static high blue for 18-hour Whole Day treatments. Intriguingly, with only six hours of R:B0:100, Morning and Evening treatments had the same high anthocyanin content as lettuce grown for 18 hours under R:B0:100. Therefore, daily blue light fraction can be reduced by using dynamic treatments to more efficiently promote growth and nutritional quality.

plant biology↗

bioRxiv2024

Thermodynamic Parameter Estimation for Modified Oligonucleotides Using Molecular Dynamics Simulations

This study investigates the thermodynamic parameters of 1,300 RNA/DNA hybrid duplexes, including both natural and chemically modified forms, using molecular dynamics (MD) simulations. Modified duplexes consist of phosphorothioate (PS) backbones and 2'-O-methoxyethyl (MOE) modifications, both commonly used in therapeutic oligonucleotides. Hybridization enthalpy and entropy were calculated from MD trajectories using molecular mechanics Poisson-Boltzmann surface area (MMPBSA) and molecular mechanics generalized Born surface area (MMGBSA) approaches. To address discrepancies with experimental data, we established empirical relationships by comparing calculated values with known experimental results of natural hybrid duplexes, then extended these relationships to the entire dataset. The corrected parameters were subsequently used to generate nearest-neighbor (NN) models, allowing for experimentally reliable melting temperature predictions. In this process, MMGBSA demonstrated superior predictive performance with high convergence and consistency for both natural and modified duplexes. Specifically, MMGBSA captured the stabilizing effects of the MOE modification with minimal bias, while MMPBSA exhibited greater variability and limited reliability. These findings highlight the potential of MMGBSA for accurate thermodynamic modeling of both natural and modified nucleic acids, providing a robust framework and experimentally meaningful insights for applications in nucleic acid-based therapeutic design and biotechnology.

biochemistry↗