Biology subjects

You, Y.

Publications and source records attributed to You, Y..

At least 19 recordsLinked to original sources

bioRxiv2026

How Much Can AI Virtual Cells Gain from Unified Perturbational Modeling?

Predicting how individual cells respond to perturbations is a central task in artificial intelligence virtual cell (AIVC) modeling. However, integrating perturbation datasets is challenging because they differ in measured gene sets, perturbation types and cellular contexts. Existing AIVC predictors are typically developed for specific settings, limiting their applicability across heterogeneous datasets. Here, we investigate the extent to which AIVC predictors can benefit from unified perturbational modeling (UPM), a novel paradigm designed to integrate heterogeneous Perturb-seq datasets into a single predictive model. To this end, we introduce three unification modules into a variational autoencoder (VAE)-based AIVC predictor to address these sources of heterogeneity: gene-channel alignment, shared perturbation encoding, and joint embedding of gene expression and identity. We systematically evaluate and validate the individual contributions of these modules. Building on UPM, we assemble a dataset of 110 million cells spanning over 22,000 genes, four perturbation types and 76 cellular contexts, and use 89.7 million cells from 71 contexts to train a unified AIVC predictor with more than one billion parameters. Across genetic perturbation benchmarks, UPM achieves the highest PES and gene-wise response correlation among the evaluated models. It also improves all four evaluation metrics over State on the five held-out Tahoe-100M cell lines and over Stack on the cytokine benchmark. UPM also improves combinatorial perturbation prediction relative to specialized predictors. In the HSC application, our model achieves a 40.0% improvement in accuracy on a downstream task involving the recapitulation of effects induced by the knockdown of hepatic stellate cell (HSC)-related targets. Our implementation is openly available at https://github.com/youlab2025/UPM.

bioinformatics↗

bioRxiv2026

Joint inference of paired dynamical gene regulatory networks reveals distinct cell-state landscapes of neutrophil reprogramming

Disease reprograms cells through changes in gene regulation, yet identifying these changes remains a major challenge. We introduce NetDes-Duo, a computational method that jointly infers transcription factor regulatory network models for two related conditions using scRNA-seq data. The networks are optimized to have minimal topological differences, while the associated ODE models recapitulate single-cell gene expression trajectories for both conditions. On synthetic benchmarks, NetDes-Duo outperformed methods that infer each network independently. NetDes-Duo was applied to neutrophil reprogramming in naive and tumor-bearing mice, and the network-simulated dynamics reproduced the observed cell state transitions. The naive landscape had two well-separated basins, whereas the tumor-bearing landscape was more continuous, with three shallower basins. Perturbation and driving simulations also identified Cebpb as a key driver of the tumor-bearing transition, consistent with emergency granulopoiesis literature. We expect NetDes-Duo to be a broadly applicable framework for uncovering the regulatory logic of disease-associated cell state transitions.

systems biology↗

bioRxiv2026

A Generative Virtual Tissue Model Enables Computational Design of Therapeutic Perturbation Strategies

Computational design has transformed many fields of engineering, where simulators can explore millions of candidate design configurations before experimental development and testing. Therapeutic design in biomedicine has resisted computational design approaches because disease progression and therapeutic response emerge from interactions among many cell types within human tissue, governed by biochemical parameters that are largely unknown and potentially unknowable. Here, we introduce the Cell Interaction Foundation Model (CIFM), a virtual tissue model that forward-simulates the transcriptional dynamics of cells in human tissue under arbitrary therapeutic conditions based upon a spatial transcriptomic seed. CIFM is a geometric graph neural network trained by self-supervised masked-transcriptome prediction on millions of cellular microenvironments spanning human tissue types and disease states; generative, auto-regressive, monte-carlo play-out, then, simulates transcriptional dynamics under combinatorial perturbations from a spatial transcriptomic seed. We validate CIFM by showing accuracy gains in gene expression prediction and imputation, disease classification, recapitulation of perturbation responses in prostate cancer models, and recovery of T cell-tumor signaling measured in cell-cell sequencing experiments. Beyond such conventional tasks, CIFM enables target identification and therapeutic design through generative tissue simulation play-outs. Analyzing over 106 single and combinatorial perturbations, CIFM designs immunotherapy strategies for cancer and autoimmune disease that exploit combinatorial manipulation of signaling pathways to induce or suppress immune activation. Broadly, CIFM shows how generative artificial intelligence methods can be applied to model emergent behavior in highly interacting biological systems, yielding new approaches to fundamental understanding of tissue behavior as well as large-scale therapeutic design.

bioinformatics↗

bioRxiv2026

Virtual-cell verification enables self-auditing AI discovery for immune rejuvenation

Artificial-intelligence agents propose drug-discovery hypotheses faster than experiments can test them, yet their conclusions are rarely verified, against the underlying biology, the predicted perturbation, or the agent's own scoring logic. We close this verification gap with an agentic framework built on three verifiers. First, PACE, a phenotype verifier, resolves immune aging into ten directionally scored, cell-type-resolved gene-set modules, selected for cross-cohort stability across four PBMC cohorts, and outperforms five established aging clocks in an independent in-house aging cohort of 434 elderly donors. Second, CellQ, a virtual-cell verifier built with multi-modal LLM, compresses each single-cell transcriptome into eight discrete tokens aligned to a language model's vocabulary through residual vector quantization; it attains state-of-the-art perturbation prediction and uniquely resolves the weak, module-level shifts that differential-expression recovery misses. Third, an Analyzer-Planner-Auditor agent verifies its own scoring logic: screening 110 compounds in primary human PBMCs, it found aged-down modules more reversible than aged-up modules and revised its objective from an equal-weight mean to a balance-constrained minimum, a self-correction that generalized to an independent 13-compound T-cell assay. By verifying its predictions and its own objective against experiment, the framework points beyond hypothesis-generating AI toward self-correcting AI scientists whose objectives could continuously evolve.

bioinformatics↗

bioRxiv2026

DigiAra Computationally Designs Plant Mutants for Resistance to Microbial Infection in Arabidopsis

Plant breeding is a resource-intensive process that requires repeated cultivation and selection across multiple generations to develop varieties with desirable traits, yet computational tools capable of supporting this process remain limited. Here, we present DigiAra, an AI-based framework for designing Arabidopsis thaliana mutants with targeted traits, particularly enhanced microbial resistance. DigiAra implements an S3 pipeline--simulation, scoring, and screening: it simulates the transcriptional effects of genetic perturbations and microbial infections, scores the predicted responses in terms of relevant traits through biological pathway analysis, and screens candidate perturbations at multiple levels. In doing so, DigiAra enables the computational exploration of the genome-wide effects of genetic perturbations and diverse microbial infections in Arabidopsis. To develop DigiAra, we address two fundamental challenges. Methodologically, we introduce a hybrid architecture that integrates local gene-level interaction modeling with global transcriptional-state modeling to predict perturbation-induced changes in the Arabidopsis transcriptional state. From a data perspective, we establish a standardized pipeline for curating, harmonizing, and processing an integrated Arabidopsis-microbe transcriptional dataset comprising 495 samples from 26 projects. As a result, DigiAra accurately predicts gene-expression changes induced by unobserved genetic perturbations and microbial infections, achieving a Pearson correlation of 0.49. Moreover, it recapitulates the general non-self response (GNSR), a 24-gene program reflecting broad transcriptional reprogramming across bacterial perturbations. In an independent study, the predicted pattern-triggered immunity pathway scores further correlate with bacterial load, with a Pearson correlation of 0.57. Lastly, we deploy DigiAra to identify 27 gene knockouts through genome-wide screening that are predicted to enhance resistance to Pseudomonas syringae pv. tomato DC3000 (Pst DC3000) while limiting growth compromise, 9 of which are supported by published studies. Together, these results establish DigiAra as an effective framework for the computational design of Arabidopsis mutants. We have made our implementation openly available at https://github.com/youlab2025/DigiAra.

bioinformatics↗

bioRxiv2026

ERK signaling integrates multiple mechanisms to drive cortical evolutionary expansion

A critical first step in human evolutionary divergence from chimpanzees and other primates is the production of greatly increased numbers of cortical pyramidal neurons during development and over evolutionary time. This expansion underpins the enlarged human cerebral cortex and, consequently, higher-order cognition and consciousness. However, the cellular and molecular mechanisms driving this progressive increase in cortical neuron number remain incompletely understood. Here we show that elevated ERK signaling in human cortical radial glia (RGs), relative to mouse RGs, arises from evolutionary changes in the developmental expression of existing shared genes. Using the Emx1-Cre line to overexpress MEK1DD, a constitutively active mutant of rat MAP2K1, we found that enhanced ERK signaling in mouse cortical RGs upregulates cAMP-PKA signaling, along with promoting self-renewal, expanding the RG pool, and accelerating cell cycle progression. Conversely, overexpression of PKA signaling in mouse cortical RGs via in utero electroporation of constitutively active PRKACA mutants (L206R or W196G) reduces ERK activity, prolongs the cell cycle, suppresses SHH-SMO and YAP/TAZ pathway activity, and blocks ependymal gliogenesis. Together, our results demonstrate that ERK and PKA signaling in cortical RGs engage a dynamic balance of synergistic and antagonistic interactions. This interplay couples evolutionarily enhanced ERK and PKA activities to increased cortical RG numbers, prolonged cell cycle progression, and an extended neurogenic period, which collectively culminate in a marked increase in neuronal production. We conclude that heightened ERK pathway activity in cortical RGs serves as a central driver of progressive neocortical expansion across mammalian evolution. Significance StatementThe human brains extraordinary cognitive abilities stem from a dramatic increase in cortical neuron numbers during evolution. This study reveals that enhanced ERK signaling in human cortical radial glia neural stem cells, relative to other primates, is a key driver of this expansion. By manipulating this pathway in mice, we demonstrate that ERK coordinates multiple mechanisms--including self-renewal, cell cycle acceleration, and a dynamic interplay with PKA signaling--to boost neuronal output. These findings identify a core molecular axis underlying mammalian cortical evolution and provide a framework for understanding how the human cortex acquired its unique size and, consequently, its remarkable complexity.

developmental biology↗

bioRxiv2026

Real Science Is Harder Than Benchmarks: Evaluating Advanced AI Frameworks on Published Studies. I. Uncertainty Quantification, ML on Therapeutic Data Commons, and Agent-Based Modeling

Artificial Intelligence (AI) frameworks for automating scientific research have shown strong performance on benchmarks, but their capacity to routinely reproduce results from multiple real-life published studies remains largely untested. We evaluated five advanced AI research frameworks (Kosmos, K-Dense, ToolUniverse, BioAgents from bio.xyz, and the AI Scientist-v2 from Sakana AI) on three real-life tasks (including two recently published papers) spanning uncertainty quantification for molecular property predictions, machine learning on Therapeutic Data Commons benchmarks, and agent-based modeling. AI frameworks demonstrated genuine strengths: generating original hypotheses, competently executing routine data acquisition and coding tasks, providing statistical measures of confidence often absent from the original papers, and producing well-formatted final reports. At the same time, our experiments revealed that real-world scientific tasks remain considerably harder than current benchmarks suggest. No AI framework matched the scope or depth of the original studies, results varied across multiple runs of the same framework with the same prompt, and we documented cases of severe hallucinations in final reports, gaps in literature coverage, and overconfident conclusions. Verification of AI outputs required substantial domain expertise. While these three tasks are only partially representative of the broader scientific landscape, they offer a starting point for developing a more rigorous methodology for evaluation of AI performance than what is currently practiced. We conclude that AI frameworks are already valuable for prototyping research directions and stress-testing completed studies, and some of the limitations documented here appear largely tractable through infrastructure improvements and continued development.

bioinformatics↗

bioRxiv2026

Carbamylome analysis reveals regulatory roles for lysine carbamylation in β-cell glycolytic and insulin-processing enzymes

In type 1 diabetes (T1D), insulin-producing {beta} cells are destroyed by an autoimmune response driven by pro-inflammatory cytokines, including interferons. {beta}-cell cytokine signaling is mediated in part by post-translational modifications, such as phosphorylation and acetylation. However, the role of other post-translational modifications in {beta}-cell cytokine signaling represents an important knowledge gap. In the context of autoimmune diseases, lysine carbamylation has gained attention for its role in pathogenesis. Here, we investigate the role of carbamylation in T1D. We found that pancreatic islet cells from the T1D model, non-obese diabetic (NOD) mice, exhibit 11% carbamylation-positive cells, whereas non-diabetic CD1 mice have only 5%. Proteomics analysis of the MIN6 insulin-producing cell line treated with a cocktail of three pro-inflammatory cytokines IFN{gamma} + IL-1{beta} + TNF identified 284 carbamylated peptides from 222 proteins impacted by the cytokine treatment. Integration of carbamylation and acetylation provided a deep view of the cytokine-regulated PTMs and potential points of interplay. A functional-enrichment analysis revealed that carbamylation was enriched in pathways related to autoimmune diseases, metabolism, DNA replication, and protein translation. Moreover, functional testing demonstrated that carbamylation inhibits the glycolytic enzyme aldolase A and the insulin-processing enzyme carboxypeptidase E, identifying a possible role for cytokine-induced {beta}-cell dysfunction. In summary, protein carbamylation is elevated in islets from NOD mice, and pro-inflammatory cytokine treatment regulates protein carbamylation in MIN6 cells. These data identify carbamylation as a potential regulatory mechanism for {beta}-cell metabolism and insulin production in the context of islet inflammation.

biochemistry↗

bioRxiv2026

RVQ-Alpha: Bridging Single-Cell Transcriptomics and Large Language Models via Discrete Tokenization and Verifiable Reinforcement Learning

Single-cell RNA sequencing yields continuous expression profiles, whereas large language models operate over discrete autoregressive sequences, leaving no shared computational interface for language-model reasoning over cell states. Existing approaches either keep cellular information outside the LLM vocabulary, consume context per listed gene, or learn reconstruction codes without gene-level grounding. We introduce RVQ-Alpha, which systematically adapts four stages of LLM training (tokenization, supervised fine-tuning, reinforcement learning, and distillation) to single-cell analysis. Multi-codebook Residual Vector Quantization (RVQ) lexicalizes each profile into a compact, hierarchical cellular alphabet in the models native token stream, while a paired decoder reconstructs the corresponding expression profile. Evidence-First supervision grounds these symbols in named genes and expression-linked evidence; Fact-Aware RLVR then penalizes contradictory claims to support auditable reasoning. Task-specific RLVR yields strong experts, but a single Mixed policy underperforms them across all four task families. To recover specialist competence in a unified model, we adopt Multi-Teacher On-Policy Distillation (MOPD), which consolidates these experts into one All-in-One checkpoint without retaining a separate policy for each task. On the CAPSTONE benchmark, a four-task suite with explicit biological shifts and deterministic ontology-aware graders, the unified checkpoint improves over Mixed on all 12 metrics and remains within 0.005 of the task-routed specialist reference on every primary metric. Overall, RVQ-Alpha provides a unified pipeline for grounded, auditable multi-task single-cell analysis.

bioinformatics↗

bioRxiv2026

CXCL-CXCR2 signaling drives cancer-endothelium interactions in SCLC metastatic seeding

Small cell lung cancer (SCLC) is a highly metastatic malignancy with tropism to the liver, yet the signals that enable organ-specific metastatic colonization remain largely undefined. During metastasis, disseminated cancer cells first encounter endothelial cells (ECs) at the vascular-tissue interface, positioning cancer-endothelium crosstalk as a key determinant of metastatic success. Defining the signaling pathways underlying this reciprocal communication may uncover actionable vulnerabilities for preventing and treating this lethal disease. Here, we uncover an EC-derived CXCL chemokine program that activates cancer-intrinsic CXCR2-RAC1 signaling as a critical mediator of SCLC liver metastasis. By integrating in vitro and in vivo models, we show that SCLC cells induce robust CXCL chemokine expression from liver ECs, which in turn enhances SCLC migration and reinforces cancer cell-EC interactions. We applied highly quantitative metastatic colony barcode sequencing coupled with individual gene inactivation to demonstrate that CXCR2 is essential for SCLC migration and liver metastatic seeding. Mechanistically, CXCL-CXCR2 signaling activates RAC1-dependent F-actin assembling to drive SCLC motility during CXCL-induced metastatic seeding. Pharmacologic inhibition of CXCR2 or RAC1 suppresses SCLC migration and prevents SCLC liver metastasis. Together, our research defined a chemokine-driven signaling circuit that governs cancer-endothelium communication during the metastatic cascade and nominate the CXCL-CXCR2-RAC1 axis as a promising therapeutic vulnerability for preventing and treating metastatic SCLC.

cancer biology↗

bioRxiv2026

Quantitative dissection of the metastatic cascade at single colony resolution

Metastasis is the leading cause of cancer-related deaths. However, the core determinants and mechanistic principles underlying the metastatic cascade remain elusive. Small cell lung cancer (SCLC) is a highly aggressive malignancy with exceptional metastatic potential and limited therapeutic options. Here, we present Metastasis Originated Barcode Sequencing (MOBA-seq), a high-throughput in vivo platform that systematically maps genetic regulators across the metastatic cascade at single-colony resolution. MOBA-seq integrates scalable barcode-based lineage tracing with a computational pipeline that quantitatively deconvolutes genotype-specific effects on metastatic seeding, dormancy, and clonal expansion across hundreds of thousands of metastatic events. Applying this approach to more than 400 candidate regulators of SCLC, we uncovered tissue-specific metastatic suppressors and universal metastatic essential genes. We identified metastatic seeding as the predominant determinant of metastasis. Comparative analysis across recipient mice of distinct genetic backgrounds further revealed that innate immune surveillance constrains metastatic progression by reducing metastatic seeding and enforcing dormancy, with additional modulation by sex and tissue context. We validated the frequently mutated gene CREBBP as a key metastasis suppressor whose loss enhances SCLC metastasis through both tumor-intrinsic and immune-modulatory mechanisms. This work establishes a scalable and quantitative platform for mapping the metastatic fitness landscape at single-colony resolution across hundreds of thousands of in vivo data points. Our approach offers a broadly applicable framework for dissecting the interactions between cancer-intrinsic and microenvironmental factors governing tumor initiation, progression, and therapeutic response.

cancer biology↗

bioRxiv2026

The role of two GLYCOALKALOID METABOLISM genes in α-tomatine biosynthesis and basal defense in tomato

Steroidal glycoalkaloids and saponins are plant cholesterol-based steroid metabolites with antimicrobial activities and potential pharmacological value. The saponin uttroside B from black nightshade (Solanum nigrum) plays an important role in defense against herbivorous insect and exhibits anti-hepatocellular carcinoma activity. The tomato (S. lycopersicum) glycoalkaloid -tomatine has been studied because of its antinutritional effects, however, its role in protecting plants from fungal pathogens remains understudied. The biosynthetic pathway of -tomatine involves multiple clustered genes designated as glycoalkaloid metabolism (GAME) genes. In this study, we generated single knockout mutants of SlGAME4 and SlGAME2 by CRISPR/Cas9-based genome editing. The SlGAME4 mutants did not accumulate glycoalkaloids but instead redirected resources towards steroidal saponin (uttroside B) synthesis. SlGAME2 mutants contained unaltered -tomatine contents indicating that the SlGAME2 gene, previously reported to catalyze the transfer of xylose to {beta}1-tomatine, is not involved in -tomatine biosynthesis. Infection assays with four fungal tomato pathogens demonstrated that the SlGAME4 mutant plants were slightly more susceptible to Botrytis cinerea, but equally susceptible to the other three fungi. Up-regulation of -tomatine-responsive genes in B. cinerea was observed during infection on SlGAME4 mutant tomato, as well as on S. nigrum suggesting that uttroside B induces a fungal transcriptional response similar to -tomatine. Furthermore, we observed that tolerance mechanisms to plant saponins mediated by glycosyl hydrolases and a glycosyltransferase contribute to virulence of B. cinerea on SlGAME4 mutant plants and S. nigrum. This indicates that also uttroside B contributes to defense against fungal pathogens and can be detoxified by B. cinerea.

plant biology↗

bioRxiv2026

PTMOverlay: A Proteomic Tool to Visualize Post-Translational Modifications Across Evolution

Evolutionary conservation has been considered a hallmark of essential basic functions in cells. Therefore, the study of evolutionarily conserved post-translational modifications (PTMs) can provide insight into their role in protein function. In this context, mass spectrometry can identify and quantify thousands of PTM sites. However, a major bottleneck lies in analyzing the large amounts of data collected by the mass spectrometer. Here we address the need for a protein sequence alignment tool for multiple PTMs across several species. We developed a tool named PTMOverlay that takes peptide identification output files and overlays PTM sites onto multiple protein sequence alignments. Examining 31 bacteria isolates, we combined their protein sequences with select PTM types, including acetylation, phosphorylation, monomethylation, dimethylation, and trimethylation. The tool revealed a variety of conserved modification sites on the bacterial central carbon metabolism. Further structural analysis revealed possible interactions between methylated arginine and lysine residues with phosphothreonine/serine sites on the homodimer interface of enolase. Overall, this tool can parse large amounts of mass spectrometry data and allows for more informed and efficient selection of sites for future studies of protein function.

systems biology↗

bioRxiv2026

Molecular basis of BET family protein inhibition by four clinical-stage small-molecule inhibitor drugs

The bromodomain and extra-terminal (BET) family proteins are critical epigenetic readers involved in various diseases, making them attractive therapeutic targets. This study elucidates the molecular basis of inhibition by four clinical-stage BET inhibitors (ZEN-3694, INCB054329, PLX51107, and INCB057643) through detailed structural comparison and analysis. Using X-ray crystallography, we determined their complex structures with the second bromodomain of human BRD2 at 1.4-1.5 [A] resolution. All inhibitors occupy the conserved acetyl-lysine binding pocket via extensive hydrophobic interactions. This hydrophobic network is further coupled with hydrogen-bond contacts for inhibitor-binding stabilization. A common feature is the formation of at least one hydrogen bond with a conserved residue of N429. Notably, INCB057643 exhibits an enhanced polar network, forming additional hydrogen bonds with residues D377 and N429, which may contribute to its improved pharmacological profile. These structures reveal a shared competitive inhibition mechanism by occluding the acetyl-lysine binding site, providing a rational basis for their transcriptional repression and guiding future inhibitor design.

biophysics↗

bioRxiv2025

Oligomeric assemblies of plant biotin carboxylase revealed by cryo-EM and cross-linking

Due to the interest in fatty acid synthesis by oilseed crops, we conducted structural studies of the biotin carboxylase (BC) subunit of the plastid acetyl-CoA carboxylase. Acetyl-CoA carboxylase catalyzes the first committed step in the fatty acid synthesis pathway and is highly regulated. Cryo-electron microscopy revealed that pennycress apo-BC forms a symmetric dimer and contains a subpopulation of a dimer-of-dimers. The domain of BC that closes over the catalytic cleft (the B-domain) appears to be dynamic, judging from the b-factors, normal mode analysis of BC structures, and its high susceptibility to acetylation. An increase in the BC concentration decreased the reactivity of the B-domain, however, suggesting structural hindrance. The partial protection of the B-domain was consistent with cross-links that formed between dimers of BC using a cross-linker cleavable in the mass spectrometer. Cross-links guided HADDOCK docking calculations suggesting a dimer of dimers of pennycress BC that is asymmetric, staggered, and tilted between dimers, with conservation in the interface. In contrast, a minimal population of a symmetric dimer of dimers with a small, non-conserved interface was observed by cryo-EM. Taken together, our structural models are the first for Brassicaceae family BC homologs and are the first from plants. These models suggest dimer interactions that might contribute to larger oligomers of BC and influence associations with other subunits of the heteromeric acetyl-CoA carboxylase.

biochemistry↗

bioRxiv2025

Multimodal Data Fusion Reveals Morpho-Genetic Variations in Human Cortical Neurons Associated with Tumor Infiltration

We introduce LetsACT (Light-Electron-Transcriptome synergistic ACTomography), a multimodal integration platform that overcomes the limitations of single-modality data acquisition and analysis of human brain cells while synergistically leveraging the strengths of each modality. Our approach enables the rapid sample preparation, cell injection, imaging, and multimodal integration of large-scale human neuronal datasets at single-cell resolution. By generating initial laser-scanning-microscopy based optical reconstruction of neuron morphologies followed by refining them using electron-microscopy derived morphological priors, we have assembled one of the largest human cortical morphology datasets to date: 8,398 neurons from 58 donors, with high cortical coverage. This platform is then applied to studying morphological impact of tumor infiltration. Pyramidal neurons in glioma-infiltrated tissues display clear volume shrinkage in somas and branches, tapering from the soma to nearby dendritic compartments. By integrating these morphological variations with spatial and bulk transcriptomic profiles, we find that glioblastoma tissues exhibit dysregulation of 15.29% of genes, including overexpression of TERT, whereas infiltrated tissues show 7.74% gene dysregulation, characterized by overexpression of tumor suppressors such as CDKN2A and TP53. Our analysis implies that pyramidal neurons observed in these infiltrated tissues may involve an active defense instead of undergoing passive apoptosis. Our finding also indicates that LetsACT establishes a valuable resource for the large-scale, comprehensive morpho-genetic analysis of human tissues.

neuroscience↗

bioRxiv2025

Biochar-induced Root Exudates Rewire the Rhizosphere Microbiome and Its Functionality

Biochar amendment is beneficial to the soil-plant system; yet the underlying mechanism is not fully understood. Integrating microfluidics and multi-omics and using wheat as a model plant, we found biochar induces differential root exudates, especially complex molecules like secondary metabolites and signaling cues, and rewires the rhizosphere microbiome, evoking a plant-beneficial rhizosphere microbiome centered by diverse plant growth promoting rhizobacteria (PGPR) known to participate in nitrogen cycling, organic matter turnover, nutrient acquisition, phytohormone metabolism, and stress resistance. Network analysis and machine learning identified high co-occurrence between specific root exudates and PGPR. qPCR, 16S rRNA gene sequencing, and untargeted soil metabolomics suggested the restructured rhizosphere microbiome may have shifted nitrogen and methane metabolism leading to reduced potential of N2O and methane emissions, and may participate in the metabolism of signaling cues and redox reactions. This study provides new insights into biochars profound impact on the soil-plant system and highlights the potential of engineering the rhizosphere through reshaping root-microbe interactions.

microbiology↗

bioRxiv2025

Building dynamical models of multi-step state transitions from single cell gene expression trajectories

Multi-step cell state transitions occur across biological processes, such as development and disease progression, yet the underlying gene regulation remains unclear. We introduce NetDes, a computational systems-biology method that infers core transcription factor (TF) regulatory networks and builds ODE-based dynamical models from single-cell gene expression trajectories. In benchmarks on synthetic trajectories with decoys and BEELINE scRNA-seq datasets, NetDes identifies regulatory interactions competitively with existing methods, and reconstructs a simulated cell-fate circuit. We applied it to time-series scRNA-seq data of iPSC-to-definitive-endoderm differentiation, epithelial-mesenchymal transition, erythropoiesis, and dendritic cell differentiation. NetDes has advantages over existing approaches in reconstructing a minimal network with a single model that reproduces observed expression dynamics and captures sequential state transitions. Network simulations predict TFs and their combinations driving each transition, recovering known master regulators, while network coarse-graining reveals the circuit logic of iPSC-to-DE differentiation. NetDes provides a general framework for mechanistic modeling of complex cell state transitions.

systems biology↗