Biology subjects

Duan, Y.

Publications and source records attributed to Duan, Y..

At least 19 recordsLinked to original sources

bioRxiv2026

Network analysis reveals converging molecular mechanisms in genetic epilepsies

Background: Rare genetic epilepsies are often difficult to diagnose and treat. They can arise from changes in many different genes, and the molecular basis linking these changes to epilepsy remains poorly understood. This complexity makes it difficult to select effective treatments and develop new therapies, leaving many patients with poorly controlled seizures and reduced quality of life Methods: To address this gap, we first identified epilepsy-associated genes and conducted a network-based analysis of biomolecular interactions in genetic epilepsy to identify converging pathways with high translational relevance and potential therapeutic targets. To run the analysis, we constructed a human brain-specific molecular interaction network by integrating all human pathways from the WikiPathways database. HotNet2 was then applied to identify highly perturbed clusters through evidence-based weight diffusion of EAGs in the human brain network. Results: We identified 2,496 epilepsy-associated genes (EAGs) from multiple resources. These genes encode coding for metabolite interconversion enzymes (20%), transporters (15%), and protein-modifying enzymes (11%). Furthermore, the network analysis identified 19 subnetworks with more than two nodes and 125 unique genes. Gene enrichment analysis revealed convergence of processes including action potential, synaptic function, nervous system development, TOR signaling, cilium organization, DNA repair, and cell fate specification. We identified 44 first-order neighbors connecting the subnetworks. By integrating drug-target interactions from DrugBank using CyTargetLinker, we highlighted potential therapeutic targets within the S4 subnetwork, associated with cerebral cortex development, as well as CREB1, a connector between subnetworks. Conclusion: Network analysis organized heterogeneous epilepsy-associated genes into biologically related subnetworks and generated hypotheses for further research.

bioinformatics↗

bioRxiv2026

A pan-cancer catalog of mutagenesis at transcription factor binding sites

Transcription factor (TF) binding can contribute to mutagenesis by altering DNA damage formation or interfering with DNA repair. Although these effects have been characterized for selected TFs and mutagenic processes, their prevalence across TFs and cancer types remains poorly understood. A systematic characterization of mutation enrichment at TF binding sites is needed to guide mechanistic studies and improve the interpretation of noncoding mutations in cancer. Here, we analyzed somatic mutations from over 4,000 tumors across 15 tissue types to generate a pan-cancer catalog of mutation patterns at binding sites for 655 TFs, organized into 276 motif clusters. We assessed mutation enrichment at single-nucleotide resolution, accounting for sequence-dependent and local variation in background mutation rates. Nearly one third of pairs with sufficient mutation counts showed significant enrichment at one or more binding site positions. Enrichment was most widespread in melanoma but was detected in all 15 tissue types, with the affected TFs and binding site positions varying across tissues. Highly expressed TFs were more likely to show enrichment than lowly expressed TFs, consistent with a role for TF binding in increased mutagenesis at the binding sites. Mutational signature analysis implicated processes beyond UV damage, including age-associated processes, deamination lesions, oxidative damage, and tobacco exposure, as potentially contributing to TF-induced mutagenesis. Together, these findings suggest that diverse interactions between TF binding, DNA damage, and DNA repair may contribute to mutation enrichment in regulatory DNA. Our pan-cancer catalog reveals position-specific enrichment of somatic mutations at TF binding sites across all 15 tissue types analyzed. By identifying specific TFs and binding site positions enriched for somatic mutations, together with potential mutagenic processes acting at these positions, our study provides a roadmap for experimental assessment of TF-induced mutagenesis and informs the interpretation and prioritization of non-coding mutations in cancer.

cancer biology↗

bioRxiv2026

Probabilistic mapping of sub-genic intolerance reveals functional and disease-critical protein regions

Different regions of genes perform distinct functions and vary in their importance to human health. Evolutionary intolerance provides a powerful means of identifying regions where disruptive mutations are under strong purifying selection, informing genetic disease discovery and variant interpretation. However, estimating intolerance in small sub-genic regions from population variation alone is underpowered and unstable. We present PRIME, a Bayesian model that stabilizes estimates of regional missense intolerance by sharing information hierarchically across regions. Importantly, PRIME produces a full joint posterior across all genes, allowing complex inferential questions that are difficult or impossible to address with existing approaches to be answered. We utilize this to identify regions enriched for pathogenic and experimentally deleterious missense variants, improve prioritization of Mendelian disease genes by focusing on their most intolerant regions, and uncover conserved patterns of purifying selection across protein families. Integrating PRIME with existing computational variant predictors improves pathogenicity prediction, demonstrating that regional missense intolerance provides complementary information for clinical variant interpretation.

genomics↗

bioRxiv2026

Haplotype-resolved genome assembly of barberry reveals structural divergence and allele-specific expression during infection by wheat stripe rust pathogen

Barberry is an ecologically and medicinally important perennial shrub and an alternate host of Puccinia striiformis f. sp. tritici (Pst), the causal agent of wheat stripe rust. However, its genome architecture and molecular responses to Pst infection remain poorly understood. Here, we generated a haplotype-resolved, chromosome-level genome assembly of Berberis aggregata using PacBio HiFi and Hi-C sequencing. The two haplotypes, Ba5A and Ba5B, were assembled into 14 pseudochromosomes each, with genome sizes of approximately 1.13 and 1.16 Gb, respectively, and showed high continuity and completeness. Comparative analyses revealed extensive divergence between the haplotypes, including widespread structural variation, substantial hemizygosity, and TE-rich, gene-poor non-alignable regions. Haplotype-specific hemizygous genes also exhibited distinct functional biases, with Ba5A enriched in immune-related processes and Ba5B associated with transport and cell-envelope functions. Dual RNA-seq across infection stages revealed stage-dependent host transcriptional reprogramming, characterized by early pathogen perception followed by stronger defense activation and photosynthesis repression, whereas Pst displayed a coordinated colonization program consistent with its biotrophic lifestyle. Allele-specific expression analysis showed that ~25-28% of genes exhibited significant haplotype-biased expression, and a subset transitioned from unbiased expression at 0 dpi to Ba5A- or Ba5B-biased expression during infection. Integration with differential expression analysis further identified infection-responsive induced-biased genes associated with defense-related processes, including responses to fungal pathogens. These findings show that haplotype divergence and dynamic allele-specific regulation jointly contribute to transcriptional responses during Pst infection, providing new insights into the genomic basis of barberry-rust pathogen interactions.

genomics↗

bioRxiv2026

Dystonia-associated TorsinA-ΔE mutation induces a gain-of-function interaction with XPO1 via its N-terminal hydrophobic segment

Childhood-onset DYT1 dystonia is a neurodevelopmental movement disorder caused by a three-base-pair deletion ({Delta}GAG; {Delta}E) in the TOR1A gene, which encodes TorsinA, a membrane-associated AAA+ (ATPase associated with diverse cellular activities) ATPase. However, the mechanisms by which the {Delta}E mutation causes neuronal dysfunction remain poorly understood. Using patient-derived neurons, we previously demonstrated that TorsinA-{Delta}E disrupts the nucleocytoplasmic transport (NCT) of both RNA and protein cargos. In the present study, proteomic analysis of induced human motor neurons revealed a markedly enhanced association between {Delta}E and exportin 1 (XPO1), a major nuclear export receptor. This aberrant association was enriched at the nuclear envelope and accompanied by impaired XPO1-mediated nuclear export. By integrating AlphaFold-based structural modeling with molecular, biochemical, and cellular analyses, we identified the N-terminal hydrophobic segment (HS) of TorsinA as a critical contributor to its interaction with XPO1. Deletion of the HS from {Delta}E reduced its association with XPO1, altered its nuclear envelope enrichment, and restored nuclear export. Moreover, expression of HS-derived peptides in patient-derived DYT1 neurons improved nuclear export, neurite outgrowth and branching, maturation-associated gene expression, and neuronal survival. Together, these findings identify an aberrant gain-of-function association between TorsinA-{Delta}E and XPO1 as a mechanism contributing to NCT dysfunction in DYT1 dystonia and establish the HS-dependent {Delta}E-XPO1 interaction as a potential therapeutic target.

neuroscience↗

bioRxiv2026

Chemically programmed multistage morphogenesis in coacervate microdroplets

Chemical reaction-driven morphogenesis enables adaptive membraneless organelles to translate biochemical inputs into coordinated structural and functional responses autonomously, but achieving non-equilibrium morphogenesis in synthetic coacervates remains challenging. Here we report an oxidation programmed multistage morphogenesis in coacervate microdroplets composed of polyethyleneimine (PEI) and sodium thioctate (ST). Coacervate microdroplets form through electrostatic complexation between PEI and ST, together with hydrophobic association among the dithiolane motifs of ST. Hydrogen peroxide converts these dithiolane motifs into more polar species, progressively weakening hydrophobic clustering, increasing hydration within the coacervate phase, and driving the microdroplets away from their initial equilibrium state. This reaction-induced compositional imbalance drives initially homogeneous microdroplets to evolve into multivacuolated intermediates, hollow structures, and finally contracted microdroplets. Experimental and simulation results support a reaction-induced secondary phase separation in which ST oxidation promotes secondary phase separation, osmotic water uptake, vacuole growth, coalescence, and shell remodeling. By recruiting glucose oxidase (GOx) into the coacervate phase to generate H2O2 in situ, we further establish an enzyme-driven route in which glucose autonomously actuates a similar sequence of multistage morphogenesis. Coupling the GOx/glucose pathway with the horseradish peroxidase/Amplex Red cascade reaction further linked glucose-triggered morphogenesis to fluorescent signal generation, enabling coacervate microdroplets to integrate biochemical sensing, structural remodeling, and optical readout. Overall, this work establishes a reaction-induced secondary phase separation for programming life-like multistage morphogenesis in membraneless microcompartments.

biochemistry↗

bioRxiv2026

Maternal diet and genetics shape the human milk metabolome

Human milk contains a diverse array of metabolites that contribute to infant nutrition, immune development, and microbial colonization. The maternal factors shaping the milk metabolome, and the relative contribution of genetics or diet vs. other factors, remain poorly understood. Here, we profiled 458 milk metabolites in 349 one-month postpartum human milk samples and integrated metabolomic data with maternal diet, clinical, transcriptomic, and genomic measurements. Maternal diet was broadly associated with milk metabolite composition, with significant correlations identified between dietary features and 323 metabolites. Coffee consumption strongly predicted milk quinic acid and 1,3-dimethyluric acid abundance, while high-fiber dietary patterns were associated with metabolites including proline-betaine and N-acetylornithine. Integration of milk transcriptomic and metabolomic data via machine learning identified biologically plausible gene-metabolite pairs, including associations between QPRT expression and quinolinic acid, and DPEP1 and cysteine-glycine dipeptide. Genome-wide association analyses identified nine study-wide significant metabolite quantitative trait loci, including novel milk-specific associations near PDE6A affecting purine metabolites and near GNE affecting free sialic acid. Comparison with plasma metabolite studies demonstrated both shared and milk-specific genetic regulation of metabolites. Finally, we found that of all tested maternal features, diet explained the largest proportion of variation in the milk metabolome. Together, these findings demonstrate that the human milk metabolome reflects both maternal exposures and mammary gland-specific biology. This work establishes a framework for understanding how genetic and environmental factors shape milk composition.

genomics↗

bioRxiv2026

Structural basis for the selective inhibition of the PI3KC3-C2 complex by Rubicon in endolysosome maturation and mitophagy

Rubicon is a negative regulator of autophagy and the endolysosomal network (ELN) and an antagonist of the class III phosphatidylinositol 3-kinase complex II (PI3KC3-C2). Inhibition of Rubicon is considered a potential means to therapeutically upregulate autophagy and the ELN to treat Parkinsons disease and other conditions characterized by autophagic and ELN dysfunction. Rubicon is specific for the UVRAG-containing PI3KC3-C2 over the purely autophagic ATG14- containing PI3KC3-C1 complex. Here, we determined the high-resolution cryo-electron microscopy structure of PI3KC3-C2 in complex with the PI3KC3-binding domain (PIKBD) of Rubicon and compared it to cryo-EM structures of unbound PI3KC3-C2 and PI3KC3-C1. Rubicon binds directly to PI3KC3-C2 only via the BARA domain of the BECN1 subunit, which is common to both C1 and C2. The selectivity of Rubicon for the PI3KC3-C2 complex over the PI3KC3-C1 complex is attributed to a conformation of the BECN1BARA domain induced by UVRAG, rather than to direct contact with UVRAG or direct antagonism by the ATG14 subunit of PI3KC3-C1. Targeted disruption of the Rubicon:PI3K3-C2 structural interface by site-directed mutations enhances mitophagic activity in human epithelial cells to levels comparable to those observed in Rubicon knockout (KO) cells. Similarly, disruption of the interaction in Rubicon-overexpressing hippocampal neurons restored lysosomal flux to wild-type levels. These data show that suppressing the function of PI3K3- C2 can fully account for the negative regulatory effects of Rubicon in the autophagy and ELN pathways. Significance StatementEndolysosome maturation and autophagosome-lysosome fusion require the production of phosphatidylinositol 3-phosphate (PI(3)P) by the class III phosphatidylinositol 3-kinase complex II (PI3KC3-C2). Rubicon is a key negative regulator of endolysosomes and autophagy that suppresses PI3KC3-C2 activity. Here, we reveal in atomistic detail how Rubicon selectively recognizes PI3KC3-C2. Disrupting the Rubicon-PI3KC3-C2 interaction restores mitophagy and enhances lysosomal activity to the same extent as Rubicon gene deletion, establishing that PI3KC3-C2 inhibition fully accounts for the biological regulatory effects of Rubicon in the autophagy and lysosome pathways.

biochemistry↗

bioRxiv2026

Quantitative profiling of intrinsic dCas9-DNA recognition reveals key determinants of guide RNA performance

CRISPR technologies based on nuclease-deactivated Cas9 (dCas9) rely on programmable DNA binding rather than DNA cleavage, yet the intrinsic DNA-recognition properties that govern optimal guide RNA (gRNA) performance remain poorly understood. Existing approaches either measure genomic occupancy in cells or infer dCas9 behavior from cleavage-based Cas9 datasets, despite DNA binding being substantially more permissive than DNA cleavage. Here we introduce TANGO (Targeted Array-based Nucleic acid-Guided Occupancy), a high-density DNA-array platform that quantitatively profiles intrinsic dCas9:gRNA binding across tens of thousands of DNA targets in a cell-free system. TANGO captures established features of dCas9 target recognition, while providing substantially greater sensitivity than prior assays. Comparison with ChIP-seq data demonstrates that intrinsic DNA-binding specificity is a major driver of genomic occupancy and reveals that chromatin accessibility modulates the intrinsic binding affinity required for dCas9 recruitment. Across CRISPRi/a guides, TANGO identifies multiple independent biochemical determinants of guide performance--including on-target affinity, mismatch tolerance, and ribonucleoprotein assembly--and flags problematic and highly promiscuous guides overlooked by current specificity metrics. Unexpectedly, some guides retain substantial guide-directed DNA binding even in the absence of a protospacer-adjacent motif (PAM), revealing an additional dimension of dCas9 specificity. Together, these results establish intrinsic DNA recognition as a quantitative and experimentally accessible determinant of dCas9 function, providing a framework for improving guide selection and enhancing the precision of CRISPR technologies.

genomics↗

bioRxiv2026

DrtA, a novel major facilitator superfamily transporter, contributes to intrinsic tolerance to the chemotherapeutic agent mitomycin C in Acinetobacter baumannii

Human-targeted non-antibiotic drugs are increasingly recognized for their intrinsic antibacterial activity, yet Gram-negative pathogens such as Acinetobacter baumannii exhibit substantial tolerance to these compounds. This tolerance is largely attributed to restricted outer membrane permeability and the activity of multidrug efflux systems. While Resistance Nodulation Division (RND) transporters have been extensively studied, the contribution of individual Major Facilitator Superfamily (MFS) transporters to non-antibiotic drug tolerance remains poorly understood. Here, we investigated H0N29_04330, designated Drug Resistance Transporter A (DrtA), a Bcr/CflA subfamily MFS transporter, to define its substrate specificity and contribution to antibiotic and non-antibiotic drug tolerance. DrtA was highly conserved across the A. calcoaceticus-baumannii complex and exhibited broad substrate specificity when heterologously expressed in an efflux-deficient Escherichia coli background, conferring resistance to benzalkonium, ethidium bromide, phenicols, and the antineoplastic agent mitomycin C. Intriguingly, drtA expression increased E. coli susceptibility to the antifolate compounds methotrexate and aminopterin, suggesting that DrtA may recognize folate-related metabolites rather than function as a dedicated antifolate transporter. In contrast, loss of drtA in its native A. baumannii host primarily impaired tolerance to mitomycin C, highlighting a context-dependent physiological role influenced by the extensive functional redundancy among A. baumannii efflux systems. Site-directed mutagenesis further identified M18 and the membrane-embedded protonatable residue D26 as critical determinants of DrtA transport activity and substrate recognition. Together with previous characterization of CraA, our findings demonstrate that Bcr/CflA subfamily MFS transporters contribute to protection against structurally diverse human-targeted compounds and expand the functional landscape of efflux-mediated intrinsic tolerance beyond conventional antibiotic resistance.

microbiology↗

bioRxiv2026

RT-nested and interfering-Primer PCR reveal prevalent isoform-specific A-to-I RNA editing in neuronal genes

BackgroundMetazoan adenosine-to-inosine (A-to-I) mRNA editing temporospatially diversifies the neuronal transcriptome and proteome. The limited read length from next-generation sequencing (NGS) constrains the quantification of the potentially differential editing levels across different splicing isoforms, restricting our understanding of the extent to which RNA editing contributes to molecular diversity and its interplay with splicing. MethodsWe employed reverse transcription nested PCR (RT-nPCR) and developed a novel interfering-Primer PCR (iPrimer PCR) technique to distinguish different transcripts of any gene. We selected multiple essential genes exhibiting RNA editing in coding sequences (CDSs) or untranslated regions (UTRs) for isoform-specific amplification and Sanger sequencing. ResultsNine different Adar isoforms together with pre-mRNA had distinct editing levels at the S>G auto-recoding site, which was predicted to have isoform-specific effects on catalytic activities. Although pre-mRNA editing might exert isoform-dependent promotion/suppression of splicing, closely located editing sites, such as those in neuronal genes qvr and stj, still exhibited high correlation in editing levels due to co-editing. iPrimer strategy further discovered differential recoding levels between the long/short 3UTR isoforms of gene jef. ConclusionsWe provide the first comprehensive solution for isoform-specific PCR amplification of any gene, enabling quantification of RNA editing level of different isoforms. Our results offer insights into how RNA editing interplays with splicing, and highlight its complicated role in expanding molecular diversity. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=79 SRC="FIGDIR/small/725286v1_ufig1.gif" ALT="Figure 1"> View larger version (17K): org.highwire.dtl.DTLVardef@3c4e61org.highwire.dtl.DTLVardef@13def88org.highwire.dtl.DTLVardef@138fa91org.highwire.dtl.DTLVardef@1ca7460_HPS_FORMAT_FIGEXP M_FIG C_FIG We developed isoform-specific PCR followed by Sanger sequencing, and achieved the quantification of differential RNA editing levels in different transcripts of a gene.

molecular biology↗

bioRxiv2026

The elusive resistome: a global comparison reveals large discrepancies among detection pipelines

Identifying antibiotic resistance genes (ARGs) from metagenomic data is critical for studying antimicrobial resistance across microbial communities and pathogens. However, there is no standardized methodology for ARG annotation. Here, we compare ten commonly used ARG detection pipelines by analysing over 270 million prokaryotic genes from the Global Microbial Gene Catalogue across 13 distinct habitats. We observed up to a 45-fold difference in the number of reported ARGs, with a mean Jaccard index of only 16% between pipelines. Pipeline selection profoundly impacted downstream biological interpretations, with drastic changes to estimates of ARG relative abundance and richness, to the characterization of pan- and core-resistomes, and to the class-level composition of the inferred resistome. ARG detection pipelines make different, defensible trade-offs, and no single approach should be treated as authoritative. Therefore, users should justify and communicate choices carefully, as our analyses show that, taken uncritically, the same data can support conflicting biological and ecological interpretations.

bioinformatics↗

bioRxiv2026

Targeted medial prefrontal cortex stimulation prevents incubation of cocaine craving and restores functional connectivity

BackgroundRelapse remains a central obstacle in the treatment of cocaine use disorder (CUD), for which no medications have received approval from the U.S. Food and Drug Administration. Transcranial magnetic stimulation (TMS) has shown promise as a potential therapeutic intervention. However, current clinical trials often rely on a "trial-and-error" approach in target selection and experimental design. We previously developed a novel TMS platform and high-density theta burst stimulation (hdTBS) technology, enabling precise, focal stimulation of the rat medial prefrontal cortex (mPFC), including the prelimbic and anterior cingulate cortices. MethodsWe applied hdTBS intervention to a well-established rat model of cocaine relapse and craving after cessation of extended access intravenous drug self-administration and assessed brain response using resting-state functional magnetic resonance imaging (fMRI). ResultsAs expected, we observed robust time-dependent increases in cocaine seeking (incubation of cocaine craving) in control rats receiving sham stimulation over 3 weeks of abstinence accompanied by a reduction in prefrontal functional connectivity. In contrast, daily sessions of hdTBS for 7 days delivered on abstinence days 14-20 prevented the emergence of the incubation effect and restored prefrontal network functional connectivity. ConclusionsThis study provides strong preclinical evidence demonstrating that precise circuit modulation of medial prefrontal subregions causally reverses both behavioral and network-level adaptations associated with relapse vulnerability. Given the clinical accessibility and established safety profile of TMS, this work provides a mechanistically grounded framework for target selection and supports the translation of focal TMS of the mPFC for relapse prevention in CUD patient. One Sentence SummaryFocal transcranial magnetic stimulation (TMS) of the mPFC using the hdTBS procedure prevents incubation of cocaine craving and restores functional connectivity.

neuroscience↗

bioRxiv2026

Cas9/sgRNA-mediated genome editing of citrus via mature tissue transformation enables both high-efficacy genome editing and early flowering

CRISPR genome editing has shown tremendous potential in genetic improvement of citrus. So far, citrus genome editing has been conducted using juvenile tissues resulting in genome-edited citrus plants that require multiple years before they can produce flowers and fruit. Here we tested whether citrus genome editing via mature tissue transformation can overcome such a hurdle. CsLOB1 is a susceptibility gene for citrus canker caused by Xanthomonas citri subsp. citri (Xcc). The transcription activator-like effector PthA4 of Xcc activates CsLOB1 by binding to the effector-binding element in its promoter (EBEpthA4-CsLOBP). In Valencia sweet orange, two CsLOB1 promoter alleles are present: TI CsLOBP, and TII CsLOBP. We specifically utilized a CRISPR/Cas9 construct (GFP-p1380N-Cas9/sgRNA:CsLOBP2) targeting EBEpthA4 in TI CsLOBP but not TII CsLOBP to test genome editing efficacy and off-target mutations. GFP-p1380N-Cas9/sgRNA:CsLOBP2 function was first validated using Xcc-facilitated agroinfiltration in Valencia leaves. The construct was subsequently introduced into Valencia mature internodal stem segments via Agrobacterium-mediated transformation, generating three independent transgenic lines (#V2, #V3 and #V5). Targeted mutations in EBEpthA4-TI CsLOBP were detected in all three lines with mutation frequencies of 100%, 21.43% and 41.94% in #V2, #V3 and #V5, respectively, while no mutations were detected in TII CsLOBP. Infection with Xcc{Delta}pthA4:dCsLOB1.3, carrying a designer TALE that specifically activates TI CsLOBP, resulted in reduced canker symptoms in #V2. Importantly, all three EBEpthA4-TI CsLOBP edited lines flowered within 15 months. In sum, these results demonstrate that CRISPR/Cas9-mediated genome modification through mature citrus transformation can achieve high genome editing efficacy and overcome the juvenility.

plant biology↗

bioRxiv2026

Task demands dynamically structure feature selection, routing, and integration in the human brain

AbstractFlexible visual categorization requires selecting task-relevant information from complex stimuli and transforming it through structured computations. Yet how task demands flexibly control these computations in the human brain remains unresolved. Dynamic faces provide a powerful test case, conveying stable identity through 3D shape and transient emotional expression through 4D movement. Using generative 4D face modelling, participant-specific MEG (N = 24), and information-theoretic analyses, we show that the brain implements a structured processing architecture. Task relevance first gates feature representations in occipital cortex, selectively sustaining relevant features while suppressing irrelevant ones. These features are then routed along pathway-specific channels--3D identity features to the ventral pathway and 4D emotion features to the lateral pathway. Finally, these inputs converge in temporal cortex, where identity and emotion features are synergistically integrated for learned identities. These results show how task demands dynamically structure feature selection, gating, routing, and integration, thereby constraining the computations that support flexible categorization.

neuroscience↗

bioRxiv2026

nELAVL phosphorylation by CDKL5 regulates inter-condensates composition and communication to promote experience-dependent maturation of the visual cortex

Mutations in Cyclin-Dependent Kinase-Like 5 (CDKL5) cause CDKL5 deficiency disorder (CDD), an X-linked neurodevelopmental condition. Through a phosphoproteomic screen, we identified the neuron-specific nELAVL family of RNA-binding proteins as direct activity-dependent substrates of CDKL5. In support of this regulatory axis, single-nuclei transcriptomics of Cdkl5 knockout (KO) cortices revealed an enriched reduction in activity-dependent mRNAs. Mechanistically, we show that nELAVL proteins undergo phase separation to form biomolecular condensates, the size of which is gated by CDKL5 phosphorylation. Loss of CDKL5 leads to enlarged nELAVL condensates, which exhibit reduced binding affinity for the target mRNA Fos, resulting in its accelerated degradation. This disruption extends to inter-condensate communication: phosphodeficient nELAVL show diminished interaction with P-bodies, which themselves become enlarged in CDD mutant iNeurons. Functionally, the absence of nELAVL phosphorylation recapitulates the deficits in experience-dependent visual function observed in Cdkl5 KO mice. Our findings establish a critical molecular mechanism by which CDKL5-mediated phosphorylation governs mRNA metabolism by tuning the properties of nELAVL condensates and their communication with other biomolecular condensates, ultimately promoting experience-dependent maturation of the visual cortex.

neuroscience↗

bioRxiv2026

Modeling gene regulatory perturbations via deep learning from high-throughput reporter assays

Assessing likely variant effects on phenotypes is of critical importance in diagnostic settings, and while much progress has been made in interpreting genic mutations based on our understanding of coding sequence, noncoding variants can be much more challenging to reliably interpret based on DNA sequence alone. High-throughput reporter assays such as STARR-seq and MPRA have shown utility in experimentally measuring regulatory effects of noncoding variants present in samples but provide no readout for variants not present in the assay inputs. However, whole-genome reporter assays provide copious data that can be used to train predictive models for prioritizing variants not directly observed in the experiment. We describe a retrainable predictive modeling framework, BlueSTARR, for this task, and present results of training several models with this framework on whole-genome STARR-seq data from two cell lines and one drug treatment. Using these models, we uncover a global signature across the human genome consistent with purifying selection against both loss-of-function and gain-of-function regulatory variants, with the latter showing a significant bias consistent with selection against gains of cis regulatory function in closed chromatin proximal to genes. By testing the model on synthetic enhancers with binding motifs for transcription factors GR and AP-1, we find that when trained on drug perturbation data, the model is able to learn distance-dependent and treatment-dependent binding patterns and their resulting reporter gene activation. These results demonstrate that lightweight, easily retrainable models such as ours have utility in probing latent signals present in novel experimental data. Finally, we find only modest differences in performance between different deep-learning architectures when trained on this single data modality, and while somewhat greater predictive accuracy can be achieved with much larger models trained at great expense on many terabytes of data, there is still copious room for improvement even for industrial strength, state-of-the-art models.

bioinformatics↗

bioRxiv2026

Long-read metagenomic sequencing reveals novel lineages and functional diversity in urban soil microbiome

City parks and other urban green spaces can bring significant benefits to the physical and mental health of city residents. However, there is limited knowledge about the microbial communities inhabiting these urban soils. Here, we applied long-read metagenomic sequencing to 58 urban soil samples from two major cities in China, enabling genome-resolved reconstruction of microbial diversity at unprecedented contiguity. We recovered 7,949 medium- and high-quality metagenome-assembled genomes, comprising 4,171 species-level genome bins, of which over 97% represent previously undescribed species. Long-read assemblies revealed extensive secondary metabolic capacity, including more than 30,000 biosynthetic gene clusters, which were highly contiguous compared with those from fragmented short-read assemblies. Beyond secondary metabolism, we uncovered over 2 million small protein families, including hundreds that are strongly enriched in the neighbourhood of defense systems and mobile genetic elements, highlighting their overlooked role in urban soils. These findings expand our understanding of the functional diversity of urban soil microbiomes and provide new insights with implications for urban public health.

bioinformatics↗