Biology subjects

Shen, H.

Publications and source records attributed to Shen, H..

At least 19 recordsLinked to original sources

bioRxiv2026

Genetic and Epigenetic Landscape of Self-Identified Hispanics/Latino in All Of Us

Human genetic studies still underrepresent many populations, limiting discovery and interpretation of medically relevant variation. Extensive admixture in self-identified Hispanic/Latino populations increases the need for representative reference data and ancestry-aware interpretation of medically relevant variation. We used long-read sequencing to analyze genetic variation, ancestry, and DNA methylation in 1,490 self-identified Hispanic/Latino participants from the All of Us Research Program. We identified 10 million novel variants and more than 100 candidate phenotype associations, with fine-mapping prioritizing structural variants over nearby single-nucleotide variants in more than half. We linked 3,329 tandem repeats to DNA methylation, resolved widespread allele-specific methylation, and identified ancestry-linked regulation across paralogous genes. These results reveal previously inaccessible genetic and regulatory variation and show the value of long-read sequencing for diverse populations.

genetics↗

bioRxiv2026

Population genomics of Actinobacillus pleuropneumoniae reveals lineage-associated K:O antigen and Apx toxin diversity

Actinobacillus pleuropneumoniae (APP) is the causative agent of porcine pleuropneumonia, causing high mortality and morbidity and substantial economic losses in the swine industry. Nineteen capsular (K) serotypes, defined by variation in capsular polysaccharide (CPS) biosynthesis loci, ten O serotypes, defined by variation in lipopolysaccharide (LPS) biosynthesis loci, characteristic Apx toxin profiles, and genomic diversity have been increasingly described; however, relationships among genomic lineage, surface antigen genotype, and Apx toxin repertoire remain incompletely resolved. Here, we analyzed 328 APP genomes, including 153 newly sequenced isolates and 175 public genomes from 21 countries. Core-genome SNP phylogeny and FastBAPS identified 23 genomic lineages. Nineteen capsular K types and 10 O-antigen types formed 26 K:O genotypes, of which 24 were restricted to a single lineage. Adjusted Wallace analysis showed greater concordance of K:O genotype with lineage than K type alone (0.881 versus 0.759), and with ApxI, ApxII and ApxIII genotype (0.949 versus 0.787). ApxIVA related sequence was detected in all genomes, while 56 complete loci revealed recurrent length variation within the N-terminus or the RTX repeat rich region and an additional divergent ApxIVA related homolog. In conclusion, our comprehensive population genomic study gives a more informative picture of APP genomics and virulence-associated diversity, providing a useful genomic basis for future diagnostic test and vaccine development.

genomics↗

bioRxiv2026

A genome-scale ORF screen for resilience factors that protect the RPE and restore vision in aged mice

Aging is a complex phenomenon, yet therapeutic discovery has largely relied on testing individual hypothesis-driven targets. Here, we performed a genome-scale open reading frame (ORF) screen for oxidative-stress resistance in retinal pigment epithelium (RPE) cells, which are vulnerable to oxidative damage during aging. The screen identified nine protective ORFs including antioxidant factors and an epithelial-to-mesenchymal transition regulator. Notably, we also identified three factors that reduce transcriptomic age including the cardiac transcription factor NKX2-5. NKX2-5 conferred robust oxidative resilience in vitro and in vivo. Domain dissection revealed that its homeodomain is dispensable for resilience, enabling development of NKX2-5{Delta}HD, which retained protective activity without detectable toxicity over 14 months. In middle-aged mice, subretinal NKX2-5{Delta}HD restored visual and electrophysiological function, and its systemic delivery improved grip strength and reduced frailty in late-aged mice. These findings highlight the value of unbiased, genome-scale gain-of-function screening, as resilience factors normally expressed in other contexts can be reengineered as potential therapeutics for age-related decline in other tissues.

genetics↗

bioRxiv2026

Loss of DNA Demethylase Enhances Arsenic Resistance Via Reducing of PIN2 Antisense Long Noncoding RNA

DNA methylation dynamics regulate diverse plant stress responses, but the mechanisms governing arsenic (As) tolerance through epigenetic pathways remain largely unknown. Here we report that Arabidopsis mutants lacking the DNA demethylase, REPRESSOR OF SILENCING (ROS1), show substantially enhanced As resistance. Integrating transcriptome and methylome data, we found that PIN-FORMED 2 (PIN2), encoding an auxin efflux carrier that also transports As out of cells, represents a key ROS1 target. In ros1, we observed increased methylation in the last exon of PIN2, a region functioning as the promoter for an antisense long non-coding RNA (lncRNA). Hypermethylation of the promoter suppresses the antisense lncRNA, which follows reduction of sense PIN2 transcription. As a result, PIN2 expression increases in ros1 mutants during As exposure, enhancing As tolerance. Analysis of ros1 pin2 double mutants confirmed that PIN2 acts downstream of ROS1 in As tolerance. Our study demonstrates an epigenetic pathway in which ROS1 regulates As tolerance by controlling PIN2 expression through antisense lncRNA-mediated transcriptional interference, offering new epigenetic strategies using plants in heavy metal contaminated environments.

plant biology↗

bioRxiv2026

Spatiotemporal atlas resolves the mesenchymal-epithelial stem cell axis governing human hair follicle aging and pathology

How adult stem cell populations coordinate tissue regeneration across physiological and pathological states remains unclear. Hair follicles contain epithelial and mesenchymal stem cells that codetermine hair regeneration. Here, we integrate a multimodal spatiotemporal atlas with functional validation to define the mesenchymal-epithelial stem cell dynamics governing human hair follicle homeostasis, aging, and androgenetic alopecia. Aging is characterized by passive depletion of hDSC-derived regenerative cues, whereas AGA involves a shift of hDSC toward an inhibitory signaling hub that promotes HFSC dysfunction. We further pinpoint hDSC-secreted PTN as a key regenerative factor that promotes hair growth and counteracts androgen-induced growth suppression. This study provides a holistic framework of adult stem cell crosstalk across physiological and pathological states, offering a mechanistic roadmap to reinstate tissue regeneration.

cell biology↗

bioRxiv2026

Molecular Editing in Protein Degradation: From a Dual NSD2/GSPT1 Degrader to an NSD2-Selective PROTAC and a GSPT1-Selective Molecular Glue

Both NSD2 and GSPT1 represent compelling targets for anti-cancer drug discovery. Herein, we carried out single-atom modification of LLC0424, an NSD2 degrader with GSPT1 neosubstrate engagement, to yield two structurally analogous molecules with divergent modes of action: 424-ND, an NSD2-selective degrader, and 424-GD, a GSPT1-selective degrader. 424-ND drove selective NSD2 degradation and suppressed androgen receptor (AR) signaling in prostate cancer cells. 424-GD induced selective degradation of GSPT1 and upregulated the integrated stress response markers ATF4 and ATF3. Biolayer interferometry revealed distinct ternary complex cooperativity profiles in the presence of NSD2 or GSPT1, which directly correlate with the observed biased degradation activity. Molecular dynamics and metadynamics simulations showed that the compounds adopt distinct low-energy conformational ensembles with different spatial orientations, likely underlying their divergent cooperativity and selectivity. These findings demonstrate that minimal structural alteration permits precision control over degrader target selectivity, while furnishing selective chemical probes for NSD2 and GSPT1.

cancer biology↗

bioRxiv2026

Dynamic Control of Prokaryotic Chromosome Ploidy Rewires Metabolic Networks to Enhance Product Biosynthesis

Building high-performance microbial cell factories requires dynamic coordination of resource allocation among cellular growth, target-product biosynthesis, and endogenous host metabolism. However, existing polyploid engineering strategies rely primarily on static manipulation of chromosome copy number. Although increasing gene dosage can enhance biosynthetic capacity, static designs cannot readily accommodate the changing metabolic demands encountered during fermentation. Here, we developed a metabolite-responsive dynamic polyploid engineering strategy that couples chromosome ploidy to the cellular metabolic state. We first constructed a high-performance L-threonine biosensor and used it to sense intracellular L-threonine levels and regulate ftsZ expression, a key cell-division gene, thereby establishing a dynamic polyploid system that requires neither exogenous inducers nor antibiotics. This system enabled engineered cells to progressively transition from polyploid to haploid during fermentation, accompanied by stage-specific remodeling of cellular physiology and metabolism. Physiological characterization revealed a marked increase in cell size and alterations in cell-envelope properties during the polyploid phase, followed by a gradual decrease in chromosome copy number as fermentation progressed. Transcriptomic and metabolomic analyses further demonstrated that dynamic ploidy transitions induced global metabolic network rewiring, remodeling the tricarboxylic acid cycle and amino acid metabolism while redirecting carbon flux toward the biosynthesis of aspartate-family amino acids. Ultimately, dynamic polyploid engineering substantially enhanced L-threonine production, enabling the engineered strain to achieve an L-threonine titer of 183.1 g/L and a yield of 0.67 g/g glucose in 5-L fed-batch fermentation without antibiotics or exogenous inducers. These findings show that dynamic regulation of chromosome ploidy can couple gene-dosage control with remodeling of cellular physiology and metabolic networks, providing a new engineering strategy to overcome the limitations of static polyploid designs and build high-performance microbial cell factories.

synthetic biology↗

bioRxiv2026

Programmable De Novo Design of Mesoporous Protein Crystal Frameworks

Three-dimensional protein crystals are ordered, porous macroscopic materials with potential applications in catalysis, biosensing, and biomedicine. However, most protein crystals are obtained by empirical screening, providing limited control over the lattice architecture, pore geometry or component composition that determine material function. Here, we present a modular strategy for the programmable design of highly porous, framework-like protein crystals using predefined protein-protein interactions. This strategy yielded over 30 distinct protein crystals, including single-component and multicomponent P213 and I213 lattices that grow to over 100 micrometers in size. Small-angle X-ray scattering and electron microscopy showed close agreement between experimental lattices and computational models. RFdiffusion-guided design generated isomorphous variants with matched lattice parameters, enabling coherent protein crystal alloys, epitaxial core-shell growth and reversible shell assembly. The designed crystals exhibit tunable mesoporous architectures, with limiting apertures of 2-18 nm, and support genetically encoded incorporation of fluorescent protein guests. These results establish a general route to programmable lattice engineering of protein crystals and position them as genetically encoded, compositionally tunable mesoporous materials.

synthetic biology↗

bioRxiv2026

MiRNA let-7a-5p Ameliorates Pulmonary Fibrosis by Suppressing TGFBR1-Mediated Endothelial-to-Mesenchymal Transition

Background Idiopathic Pulmonary Fibrosis (IPF) is a fatal chronic lung disease with limited therapeutic options. While alveolar epithelial injury and fibroblast activation are well-studied, endothelial-mesenchymal transition (EndoMT) is emerging as a critical pathogenic mechanism. The regulatory role of exosomal miRNAs in pulmonary fibrosis remains unclear. This study investigates serum exosomal miRNAs, particularly let-7a-5p, in modulating EndoMT during the onset of pulmonary fibrosis. Methods Clinical cohorts of IPF patients and healthy controls were enrolled. Serum exosomal miRNAs were profiled, followed by differential expression and functional enrichment analyses. In vitro experiments involved human pulmonary artery endothelial cells (HPAECs) transfected with let-7a-5p mimic or inhibitor. Dual-luciferase reporter assays confirmed the binding between let-7a-5p and TGFBR1. HPAECs were co-cultured with lung epithelial cells to examine paracrine signaling. In vivo studies used a bleomycin-induced mouse model with let-7a-5p agomir administration. Assessments included histopathological staining, hydroxyproline content, Western blot, qPCR, micro-CT, and pulmonary function tests. Results Let-7a-5p was significantly downregulated in serum exosomes from IPF patients, correlating with clinical indicators. Mechanistically, let-7a-5p directly bound the TGFBR1 3'UTR to inhibit its expression. Inhibition of let-7a-5p upregulated -SMA, FN1, smad2/3 phosphorylation, and collagen I, while downregulating CD31 and VE-cadherin. Therapeutically, let-7a-5p mimic reversed bleomycin-induced EndoMT and suppressed epithelial-mesenchymal transition (EMT) via paracrine signaling. Mice administered agomir showed reduced fibrosis, improved lung function, and suppressed TGF-{beta}/Smad signaling. Conclusion Serum exosomal let-7a-5p suppresses pulmonary fibrosis by targeting TGFBR1 to inhibit EndoMT. Its downregulation in IPF patients correlates with disease progression, highlighting its biomarker potential.

molecular biology↗

bioRxiv2026

Eco-evolutionary feedbacks generate bistability in population persistence under gradual environmental change

Understanding how populations persist in gradually deteriorating environments through evolution is a central question in ecology and evolutionary biology. Previous studies have primarily focused on identifying the critical rate of environmental change beyond which extinction is certain. However, the existence of a viable equilibrium when the rate is below the threshold does not guarantee that a population can survive the transient dynamics to reach it. Using a quantitative genetic model that explicitly incorporates feedback among population size, genetic variance, and mean trait evolution, we show that population persistence can exhibit bistability when the rate of environmental change is below the extinction threshold. Specifically, extinction still occurs if the initial population size and genetic variance fall below a critical level. The initial state also influences the eco-evolutionary dynamics, such that a temporary increase or decline in population size and/or genetic variance does not necessarily predict the ultimate fate of the population. Therefore, in addition to estimating the critical rate of environmental change for extinction, characterizing current population size, genetic variation, and the degree of maladaptation may improve predictions of extinction risk in deteriorating environments.

evolutionary biology↗

bioRxiv2026

ROS Impair Mitophagy via PARylation of PINK1

Mitophagy is the process of selective autophagic clearance of damaged mitochondria and is closely implicated in neurodegenerative disease. PTEN-induced kinase 1 (PINK1) and a RBR E3 ubiquitin-protein ligase (Parkin) constitute a positive feedback loop in mitophagy initiation. It is known that reactive oxygen species (ROS) modulate mitophagy, while the exact regulatory mechanism remains largely elusive. Here, we found that exogenously applied ROS effectively block mitophagy induced by acute mitochondrial damage agents, which could be reversed by antioxidants. Mechanistically, ROS activate poly(ADP-ribose) polymerase 1 (PARP1), and suppression of PARP1 eliminates the inhibitory effect of ROS on mitophagy. Notably, PARP1 directly interacts with PINK1 and mediates its PARylation at residue E417, thereby negatively regulating PINK1 function. Collectively, our study identifies PARylation as a new form of post-translational modification of PINK1 and reveals a novel mechanism underlying the regulatory role of ROS in mitophagy by PARP1 activation and PARylation of PINK1. In briefGao et al. demonstrate that exogenous ROS inhibit mitophagy. Mechanistically, ROS activate PARP1, which mediates PARylation of PINK1, a central regulator of mitophagy, leading to its functional impairment. This study reveals a novel regulatory mechanism of ROS on mitophagy through PARP1 activation and identifies PARylation as a novel form of post-translational modification of PINK1. HighlightsO_LIROS block PINK1-Parkin-mediated mitophagy. C_LIO_LIROS activate PARP1. C_LIO_LIPARP1 suppression eliminates the inhibitory effect of ROS on mitophagy. C_LIO_LIPARylation of PINK1 by PARP1 impairs its activity and mitophagy. C_LI

Cell Biology↗

bioRxiv2026

Multisensory integration of a host metabolite coordinates bacterial predation of macrophages

The crisis of antimicrobial resistance demands new strategies that circumvent conventional killing. Here, we show that Yersinia pseudotuberculosis intercepts host mesifurane, a mimic of quorum-sensing signal AI-2. Rather than acting through a single receptor, mesifurane is directly and independently sensed by three receptor classes, YcrA, NarX, and CusR, that converge on a "Sense, Arm, Recruit" program. YcrA drives chemotaxis toward host cells; NarX-NarP licenses T6SS-mediated killing; and CusR amplifies AI-2 production to recruit more bacteria. YcrA is conserved across Gram-negative pathogens, creating a therapeutic vulnerability. Suramin sodium, a repurposed drug blocking YcrA, disarms this circuit without inhibiting growth. In murine sepsis, suramin sodium conferred 80-90% survival against WHO critical-priority pathogens, including carbapenem-resistant A. baumannii, E. coli, and P. aeruginosa, even when delayed 8 hours post-infection, without selecting for resistance. These findings establish that disrupting sensory perception, rather than viability, provides an anti-infective strategy bypassing resistance while imposing minimal selective pressure.

microbiology↗

bioRxiv2026

Intron 13 retention expands the ADAR1 isoform repertoire to rewire PKR signaling and promote tumorigenesis

For decades, the biology of ADAR1 has been framed around two major isoforms: the nuclear-enriched p110 and the predominantly cytoplasmic p150. Here, we reveal an unexpected repertoire expansion of ADAR1 isoforms that is driven by intron retention. Specifically, intron 13 (I13) can be retained in ADAR1 transcripts, conferred by an evolutionarily conserved weaker 5SS. In addition, we found that the I13 retention (I13R) is negatively autoregulated by ADAR1 through antagonizing the binding of hnRNPA1 to I13 in an editing-independent manner. Despite being sensitive to nonsense-mediated decay, I13R generates two previously uncharacterized truncated isoforms - p90 and p130, that both lack the C-terminal portion of the deaminase domain. Intriguingly, ADAR1p90 - a derivative of the canonical nuclear-enriched p110 isoform - is predominantly cytoplasmic, that effectively represses PKR and eIF2 activation through the sequestration of immunogenic double-stranded RNA (dsRNA) substrates. Furthermore, in a colorectal cancer (CRC) cohort, p90 levels are increased in most tumors relative to matched normal tissues that positively correlates with hnRNPA1 expression. Functionally, xenografts expressing ADAR1p90 grow significantly faster and larger than those expressing ADAR1p110, indicating enhanced tumorigenic potential. These findings revise the canonical view of ADAR1 isoforms, demonstrating that intron retention can generate alternative isoforms with augmented functions that tumors readily exploit. One Sentence SummaryThe conserved retention of I13 in ADAR1 transcripts gives rise to previously uncharacterized C-terminal truncated cytoplasmic ADAR1 isoforms, p90 and p130, which, although devoid of catalytic activity, sequester immunogenic dsRNA substrates, thereby preventing PKR binding and downstream activation of eIF2.

cell biology↗

bioRxiv2026

Survivin inhibition induces PD-L1 expression through cGAS stabilization and attenuates antitumor immunity

BackgroundSurvivin is a mitotic regulator frequently overexpressed in human cancers and an attractive therapeutic target. However, how Survivin inhibition influences tumor immune regulation remains incompletely understood. This study aimed to investigate whether Survivin inhibition modulates antitumor immunity and to elucidate the underlying mechanisms. MethodsProgrammed death-ligand 1 (PD-L1) expression was evaluated in multiple tumor cell lines following pharmacological or genetic inhibition of Survivin. Mechanistic studies included RNA sequencing, immunoblotting, flow cytometry, cGAS knockdown, and NF-{kappa}B inhibition. Immune profiling was performed using CD8 T-cell cytotoxicity assays, mass cytometry, flow cytometry, immunofluorescence and single-cell RNA sequencing. Clinical relevance was assessed using patient tumor specimens and public immunotherapy cohorts. ResultsSurvivin inhibition, either by YM155 treatment or genetic depletion, increased PD-L1 expression at both mRNA and protein levels in tumor cells. Mechanistically, Survivin inhibition stabilized cyclic GMP - AMP synthase (cGAS) by reducing its ubiquitination and activated NF-{kappa}B signaling, thereby promoting transcriptional upregulation of PD-L1. Functionally, the induced PD-L1 enhanced PD-1 engagement and suppressed CD8 T-cell cytotoxicity, promoting immune evasion. In immunocompetent ovarian cancer models, pharmacological inhibition of Survivin increased PD-L1 expression in tumor and immune compartments and attenuated cytotoxic immune activity. PD-L1 blockade restored antitumor immunity and significantly enhanced the therapeutic efficacy of Survivin inhibition. In addition, analyses of patient samples and public single-cell datasets revealed an inverse association between Survivin and PD-L1 expression, and high Survivin expression was associated with reduced benefit from PD-1/PD-L1 blockade. ConclusionsThese findings identify a Survivin-cGAS-PD-L1 axis linking mitotic stress to immune suppression and provide a mechanistic rationale for combining Survivin-targeted therapy with immune checkpoint blockade.

cancer biology↗

bioRxiv2026

Top-down regulation of ingestive behavior fragmentation

AbstractIn natural environments, animals rarely feed continuously to satiation; instead, feeding occurs in brief bouts separated by pauses. This fragmentation is thought to balance internal drives with external demands, yet its underlying neural mechanisms remain unclear. By combining bidirectional neural activity mapping and behavioral phenotyping, we identify a projection from the dorsal subiculum (dSub) of the hippocampus to the mammillary body (MB) as a key regulator of this fragmentation. Activity along the dSub-MB pathway tracks and gates the duration of individual feeding bouts, independent of homeostatic state. A simple bistable attractor model captures both dSub-MB neural dynamics and associated behaviors across optogenetic and behavioral perturbations. Together, these findings identify a top-down circuit mechanism that implements action selection in a naturalistic setting.

neuroscience↗

bioRxiv2026

Barrier Immune Memory is Promoted by Intestinal Epithelial Cell Presentation of Injected Bacterial Antigens

The contributions of antigen compartmentalization to recognition differences between CD4 and CD8 T cells have long been appreciated, but little is known of how subcellular localization of different antigens expressed by a single pathogen impacts T cell immunity. By tracking a clonal CD4 T cell response to its cognate epitope shuttled between different virulence proteins of the enteropathogenic bacterium, Citrobacter rodentium (Cr), we find a remarkable bias in the magnitude and quality of the response contingent on whether antigen remains bacterially associated or is introduced into intestinal epithelial cells colonized by the bacterium. Only proteins injected into the cytosol of colonocytes via the type 3 secretion system (T3SS) of Cr were found to recruit robust antigen-specific T cell responses to the infected mucosa and give rise to CD4 resident memory T (Trm) cells that populate the mucosal epithelium--and this required direct presentation of these antigens by infected epithelial cells. Single-cell transcriptomic analyses revealed that sustained, bidirectional epithelial-T cell communication was required both to elicit epithelial barrier-protective T cell help and to promote transcriptional networks that program a tissue-residency rather than central memory fate. These results establish a central role for antigen presentation by non-professional APCs in controlling memory fate decisions by CD4 T cells, with important implications for development of successful mucosal vaccines.

immunology↗

bioRxiv2026

Gq-pathway activation in hippocampal CA1 astrocytes rescues ischemia-induced memory deficits and synaptic plasticity

More than 30% of stroke survivors develop post-stroke cognitive impairment (PSCI), for which current neuron-centric therapies not only lack cellular specificity but also carry risks of adverse effects such as epilepsy. Here, we explore the therapeutic potential of astrocytes by chemogenetically activating Gq signaling in hippocampal astrocytes, which rescues memory deficits and synaptic plasticity impairments in a mouse model of ischemic stroke. Gq activation restores dendritic complexity, spine density, and long-term potentiation in hippocampal CA1 neurons. Fiber photometry further reveals that astrocytic Ca{superscript 2} signals precede neuronal activity by 600 ms during novel environment exploration, indicating that astrocytes prime memory encoding. In contrast, Gi pathway activation induces pathological neuronal hyperactivity without cognitive improvement. These findings establish that astrocytes regulate post-stroke recovery through pathway-specific calcium signaling and uncover a previously unknown temporal hierarchy in astrocyte-neuron communication during memory processing, offering a new glia-targeted strategy to overcome the limitations of current neuromodulation approaches for PSCI.

neuroscience↗

bioRxiv2026

Taxonomy-aware, disorder-matched benchmarking of phase-separating protein predictors

BackgroundBiomolecular condensates formed via liquid-liquid phase separation (LLPS) play vital roles in cellular organization and function. Computational prediction of phase-separating proteins (PSPs) is increasingly used to prioritize candidates at proteome scale, making robust, well-designed benchmarks essential for fair evaluation and iterative improvement of PSP predictors. ResultsWe first show that a recently released PSP benchmark is substantially confounded by the imbalances in taxonomic origin and intrinsic-disorder compositions between positive and negative sets, allowing predictors to achieve high apparent performance by exploiting non-LLPS shortcuts and obscuring their true ability to distinguish PSPs. To minimize these effects, we construct a taxonomy-aware, disorder-matched PSP benchmark. Using this benchmark, we find that absolute sequence and biophysical feature values of PSPs differ markedly across taxa, whereas LLPS-associated feature shifts relative to taxon-specific proteome backgrounds are comparatively conserved. Benchmarking twenty PSP predictors under this framework reveals pronounced taxon-dependent variation in performance. Moreover, PSPs lacking IDRs consistently constitute a more challenging regime across methods, motivating routine disorder-stratified evaluation. ConclusionsOur taxonomy-aware, disorder-matched benchmarking framework reduces shortcut-driven biases, enables more interpretable evaluation of PSP predictors, and provides guidance for developing models that capture transferable LLPS-associated signals rather than dataset- or taxon-specific shortcuts.

bioinformatics↗