Biology subjects

Liu, H.

Publications and source records attributed to Liu, H..

At least 73 records · Page 4Linked to original sources

bioRxiv2026

Promoter architecture diversifies ParR-dependent transcription in Pseudomonas aeruginosa

Two-component systems convert environmental signals into transcriptional responses, but how one response-regulator phosphorylation state produces different outputs at different promoters remains unclear. Here we show that promoter architecture determines how the ParR response regulator in Pseudomonas aeruginosa interprets phosphorylation. Phenotypic, transcriptomic, biochemical and promoter-engineering analyses showed that phosphorylation of the conserved ParR receiver residue Asp57 lowered DNA-occupancy thresholds across target promoters. Individual promoters, however, converted this shared increase in binding into activation, repression or phosphorylation-dependent sign switching. Chromosomally D57 ParR variants reproduced these behaviours at endogenous loci and altered antibiotic susceptibility, biofilm formation and virulence-associated phenotypes. Mapping and engineering of representative promoters identified a two-tier cis-regulatory code: half-site sequence compatibility determines productive ParR engagement, whereas spacer length determines regulatory sign and magnitude. Spacer swaps were sufficient to reprogram promoter logic between regulatory modes. These findings separate regulator state from promoter decoding and show how bacterial promoter architecture can diversify the outputs of a shared phosphorylation signal. Graphical abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=142 SRC="FIGDIR/small/738673v1_ufig1.gif" ALT="Figure 1"> View larger version (54K): org.highwire.dtl.DTLVardef@1a0a26forg.highwire.dtl.DTLVardef@10836a8org.highwire.dtl.DTLVardef@3c6cd8org.highwire.dtl.DTLVardef@10dfefb_HPS_FORMAT_FIGEXP M_FIG C_FIG HighlightsParR coordinates antibiotic resistance, biofilm formation and virulence-linked phenotypes in P. aeruginosa. Promoters convert ParR phosphorylation into activation, repression or sign switching. Half-site compatibility and spacer length form a cis-regulatory code for ParR output.

biochemistry↗

bioRxiv2026

OTTR-CLASH: improved biochemical and bioinformatic identification of Argonaute 2-mediated microRNA-target RNA interactions

Various methods have detected miRNA-target interactions via immunoprecipitation of UV-crosslinked Argonaute ribonucleoprotein complexes, followed by intermolecular ligation of bound miRNAs to target strands, forming chimeric RNAs. To date, these methods have relied on conventional viral reverse transcriptases (RTs) to generate cDNAs for sequencing. However, crosslinked RNAs often retain adducts after purification, which can make them poor templates for viral RTs. Here, we adapted OTTR (Ordered Two-Template Relay) techniques to generate cDNAs from Ago2-bound RNAs. OTTR makes use of a modified retroelement-encoded RT, which is strongly processive even on templates with modifications or adducts. We show that this "OTTR-CLASH" method increases the frequency of generating chimeric RNAs compared to previous methods. We also developed an improved bioinformatic pipeline for analysis of these data, and we use this to catalog miRNA-target interactions not previously described in the literature. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=147 HEIGHT=200 SRC="FIGDIR/small/738487v1_ufig1.gif" ALT="Figure 1"> View larger version (24K): org.highwire.dtl.DTLVardef@16d62f8org.highwire.dtl.DTLVardef@7c8da4org.highwire.dtl.DTLVardef@13727fborg.highwire.dtl.DTLVardef@21f443_HPS_FORMAT_FIGEXP M_FIG C_FIG

molecular biology↗

bioRxiv2026

MAERM: Predicting Enzyme-Reaction Matching Relationships with a Mixed-Attention Model

Harnessing enzyme specificity requires a thorough understanding of enzyme promiscuity, which determines enzymes catalytic scope; however, measuring this scope still relies heavily on labor-intensive analytical approaches. While data-driven approaches have emerged to predict the catalytic scope of enzymes, these methods continue to face challenges such as restricted datasets and insufficient integration of enzyme structural information and reaction transformations. Here, we introduce MAERM, an innovative mixed-attention model designed to predict enzyme-reaction matching relationships. Built on our MAERM-DB, a dataset with broad coverage of validated and chemoenzymatic catalysis data, MAERM utilizes a local-global attention module to integrate multimodal enzyme information with fine-grained reaction representations, thereby predicting enzyme-reaction matching probabilities. Results show that MAERM consistently outperforms all baselines, with an average F1-score of 0.984. Notably, on challenging test samples with less than 40% sequence identity to the training set, MAERM outperforms the second-ranked model by 5.9% in F1-score. In addition, MAERM achieves the highest top-10 success rate of 51.7% on Enzyme-405 and the highest balanced accuracy of 0.697 on BioCat-547, further supporting its generalizability in enzyme screening and chemoenzymatic catalysis. Finally, MAERM can serve as an efficient scoring module. When integrated with ProteinMPNN, MAERM has successfully guided novel enzyme design for two carbonyl reduction reactions, resulting in enhanced catalytic potential for the native substrate and demonstrating broad compatibility. Overall, MAERM has the potential to reduce the experimental cost of measuring enzymes catalytic scope, facilitate enzyme design, and ultimately accelerate the design-build-test-learn cycle in enzyme engineering.

bioinformatics↗

bioRxiv2026

An allosteric pocket in KV1.3 defines a distinct chemical space for immunomodulator design

KV1.3 is a validated autoimmune drug target, yet all disclosed inhibitors converge on a narrow set of conserved binding sites, suffering from poor selectivity or clinical failure. Whether KV1.3 harbors a distinctive, druggable pocket amenable to selective targeting has remained unknown. Here we report MPiN, a highly selective, state-dependent KV1.3 inhibitor with in vivo efficacy in a mouse psoriasis model, which engages a previously uncharacterized extracellular allosteric pocket framed by the PP1-PP2 turret loops, the pore helix and the outer S5/S6 helices, acting through a bidirectional pore-to-sensor coupling that simultaneously constricts the selectivity filter and facilitates the voltage sensor toward activation. Unexpectedly, despite the extensive structural conservation of this pocket across KV1 paralogs, subtype selectivity is dictated by the peripheral residues G427 and H451, which define pocket geometry without directly contacting the ligand, thereby establishing a geometry-driven "non-contact selectivity" mechanism. By opening an unrecognized, structurally distinct chemical space on KV1.3 and redefining how selectivity is achieved within a conserved channel family, this work lays the structural and conceptual foundation for rational, structure-guided design of next-generation KV1.3 immunomodulators.

biochemistry↗

bioRxiv2026

Designing Fidelity of CRISPR-Cas Endonucleases by Kinetic Insights

Finding high-fidelity CRISPR-Cas variants is critical for both the precision of in vitro DNA detection and the safety of in vivo gene-editing therapeutics. However, the large size of the Cas enzyme and the distinct selection criteria between its natural evolution and clinical practice lead to extensive experimental trials and limited success rate for de novo design, directed evolution, protein language model (PLM)-based filtering. Here, we present a PLM-assisted physics-driven approach that utilizes atomistic molecular dynamics simulations and automated path searching to efficiently obtain the complete kinetic insights, including the transition state structures, for the conformational changes of Cas before DNA cleavage. We show that these kinetic insights can pinpoint a few fidelity-diminishing protein residues during the early stage of target recognition, and have led to SpyCas9 and FnCas12a variants with ultra-high fidelity surpassing previously reported counterparts at minimal cost of a few wet-lab trials.

biophysics↗

bioRxiv2026

TLR7 Signaling Regulates Embryonic Hematopoietic Stem Cell Development by Sensing microRNA-146a in Vertebrates

The specification of hematopoietic stem cells (HSCs) is tightly regulated by multiple transcription factors and signaling pathways. Inflammatory signaling is pivotal for embryonic HSC development, but the mechanisms that activate it in vivo remain poorly understood. Here, we show that Toll-like receptor 7 (TLR7) is essential for the emergence of embryonic HSC in both zebrafish and mouse embryos. TLR7 deficiency reduces HSC numbers but not primitive or definitive progenitors. Conversely, the TLR7 agonist R848 enhances embryonic HSC development. Mechanistically, TLR7 signaling acts through interferon regulatory factor 5 (IRF5) to induce the expression of inflammatory cytokines, which subsequently activate Notch signaling to promote HSC emergence through a non-cell-autonomous mechanism. Notably, we identify microRNA-146a (miR-146a) as a potential endogenous activator of TLR7, inducing inflammatory signaling and promoting HSC development. Pharmacological treatment with miR-146a significantly increases HSC numbers in zebrafish embryos. Together, our findings reveal a crucial role for miR-146a-TLR7-IRF5 signaling axis in HSC emergence, providing insights into the endogenous factors that drive tonic inflammatory signaling during normal hematopoiesis and suggesting the translational potential of TLR7 agonists and miR-146a for stem-cell-based therapeutics. Significance StatementThe embryonic origin of hematopoietic stem cells (HSCs) requires inflammatory signals, but the endogenous factor that triggers this process remains elusive. We identify microRNA-146a (miR-146a) as a natural activator of Toll-like Receptor 7 (TLR7) signaling, which is essential for HSC emergence. This miR-146a-TLR7 axis functions through IRF5 and inflammatory cytokines to activate the Notch signaling, specifically promoting embryonic HSC development. Our work addresses the critical question of endogenous ligands that mediate tonic inflammatory signaling in normal hematopoiesis, uncovers novel crosstalk between miRNAs and innate immunity in HSC specification, and identifies promising candidates for stem cell-based therapeutics.

developmental biology↗

bioRxiv2026

DELPHAI predicts heterogeneous perturbation responses with learned cell fitness and gene-space retrieval

Current perturbation response modelling in single-cell transcriptomics assumes conserved cell mass and loses gene expression information to latent-space decoding. We propose DELPHAI, training a fitness network and an optimal transport network jointly without biological priors, and during inference applying a fitness-gated transport with a direct gene-space retrieval. Demonstrated across two benchmark frameworks, DELPHAI ranks first in predicting differentially expressed genes, while revealing which cell lineages a perturbation depletes.

bioinformatics↗

bioRxiv2026

Rapid immunostaining and high-resolution three-dimensional light-sheet microscopy of intact calcified tissues

High-resolution 3D imaging is an important strategy for visualizing and analysing complex skeletal tissue architecture and the bone marrow microenvironment. However, multicolor immunolabeling and imaging of intact skeletal tissues are technologically challenging. The current immunolabeling and clearing methods for intact skeletal elements are very limited, time-consuming and generate low-resolution data or depend on the use of reporter mice. Here, we describe a protocol for efficient clearing and immunolabeling of intact calcified tissues that enables superfast, single-cell resolution, and quantitative 3D light-sheet imaging of intact skeletal elements and teeth. A key aspect of our protocol is the addition of a collagenase digestion step after fixation and decalcification. This step enhances antibody penetration, resulting in deep, comprehensive staining throughout immunostained bones and other calcified tissues. The protocol includes soft tissue removal, fixation, decalcification, bone dehydration, and bleaching, followed by antigen retrieval and permeabilization before the collagenase digestion step. This procedure is performed to prepare the samples for the tissue clearing process that improves bone tissue transparency prior to light-sheet imaging. The entire protocol, from bone collection to image analysis and quantification, takes about 4 days to complete, thus offering significant improvements over previous methods. This protocol is broadly applicable to the visualization of bone microstructure, bone marrow analysis, vascular and neural network mapping, and the study of signaling molecules in bone development and growth. The protocol requires experience with standard tissue processing and immunostaining techniques, and prior experience in tissue clearing and light-sheet imaging is beneficial but not essential. Key pointsO_LIA protocol for efficient clearing and immunolabeling of intact calcified tissues that enables superfast, high-resolution, and quantitative 3D imaging of various intact bones and teeth. C_LIO_LIThe entire protocol takes only 4 days to complete the comprehensive staining and perfect transparency throughout the intact bones, offering significant improvements over previous methods. C_LI Key referencesBiswas, L. et al. Cell 186, 382-397.e24 (2023): https://doi.org/10.1016/j.cell.2022.12.031

cell biology↗

bioRxiv2026

Protocol for standardized minimally invasive mouse models of bisphosphonate-related and radiation-induced jaw osteonecrosis

This protocol describes a standardized and reproducible minimally invasive approach for establishing mouse models of bisphosphonate-related osteonecrosis of the jaw (BRONJ) and osteoradionecrosis of the jaw (ORNJ). The method combines a unified low-trauma oral surgical procedure with disease-specific injury induction strategies to generate robust and clinically relevant models of jaw osteonecrosis. For BRONJ, systemic zoledronic acid administration is coupled with mandibular first molar extraction using tape-assisted mouth opening and customized bent micro-forceps, minimizing soft tissue damage and reducing procedural variability. For ORNJ, a customized lead-shielding platform enables precise, noninvasive mandible-targeted irradiation, producing reproducible bone injury while limiting off-target radiation exposure. Together, these complementary models provide a consistent and minimally invasive framework for investigating jaw osteonecrosis arising from distinct etiologies. The protocol supports comprehensive downstream analyses, including micro-computed tomography, histology, and immunofluorescence, and facilitates mechanistic studies of disease pathogenesis, bone regeneration, and therapeutic intervention.

pathology↗

bioRxiv2026

AI-enabled Spatial Profiling of Circulating Tumor-Immune Ecosystems Predicts Patient Outcomes Across Cancers

Circulating tumor cells (CTCs) and immune cells form dynamic multicellular ecosystems in blood, but their spatial organization and clinical relevance have not been systematically characterized. We developed the Cell and Cluster Identification Program (CCIP), an artificial intelligence-based framework that analyzes routine multiplex immunofluorescence blood scans to segment cells, identify CTCs and five immune lineages with high accuracy, and quantify multicellular clusters and tumor-immune interactions. Applying CCIP to 2,693 blood scans from 1,399 patients, we profiled over 60 million cells (>7 million multi-cell clusters) and linked imaging-derived features to patient outcomes. Correlated with circulating-tumor DNA mutation burdens, a 14-feature image model predicted overall survival in breast cancer, outperformed clinicopathologic variables and CTC enumeration, and generalized to prostate cancer. Prognostic imaging signatures were also associated with therapy response-related progression-free survival as well as with single-cell RNA sequencing-derived immune suppression states, connecting circulating tumor-immune architecture with systemic immune dysfunction.

cancer biology↗

bioRxiv2026

Telomerase RNA regulates the epigenome primed for human lineage commitment

Telomerase RNA (TERC) is known as the essential template for telomere elongation. Here, we report an unexpected role for TERC in regulating chromatin accessibility, which is primed for determining the cell fate of human embryonic stem cells (hESCs). TERC-deficient hESCs retain critical markers for pluripotency but fail to undergo lineage differentiation as shown by standard in vivo teratoma formation as well as in vitro differentiation assays, which is consistent with repressed transcription during differentiation into the three germ lineages. Notably, transient re-introduction of TERC into TERC-deficient hESCs rescued lineage differentiation capacity without restoring telomere length. TERC binds to the promoters and enhancers of developmental genes marked by H3K27ac to maintain an open chromatin state. Loss of TERC reduces H3K27ac deposition and decreases chromatin accessibility through the remodelling of three-dimensional genome organization, including TAD boundary insulation and compartment switching. Collectively, our findings reveal that TERC is a chromatin-associated noncoding RNA that regulates the epigenomic architecture that governs cell fate for lineage commitment during development.

cell biology↗

bioRxiv2026

Preclinical translation of Neurofibromatosis type 1 (NF1) exon 17 skipping using targeted U7-SnRNA packaged into engineered AAV serotypes.

To facilitate the translation of NF1 exon 17 skipping as a mutation-specific therapy for Neurofibromatosis type 1 into in vivo testing, we have continued to develop more efficient antisense oligonucleotides (ASOs), humanized mouse models, and explored multiple delivery platforms including an adeno-associated virus (AAV)-U7-SnRNA vector approach. We evaluated both biodistribution and exon skipping efficacy of a U7-SnRNA targeting NF1 exon 17 with an SFFV-driven cassette containing T2A-linked Luciferase (Luc) and eGFP packaged in AAV-9, AAV-F and AAV-B1 capsids. We show that AAV-F is superior to AAV-9 and AAV-B1 for mouse brain delivery based on DNA transduction, GFP expression, and luciferase activity, but AAV-B1 delivers 2-4 fold more to sciatic nerve (SCN). In terms of exon skipping, AAV-F appears to induce the most skipping in liver and optic nerve (ON), while AAV-B1 mediates highest skipping in the liver, SCN, and ON. The identification of AAV serotypes that allow efficient transduction and delivery of transgenes to the mouse CNS and PNS is impactful for preclinical research in murine models of other diseases. Furthermore, this is both the first report of NF1 exon skipping efficacy in vivo and the first successful application of an U7-SnRNA for the restoration of functional neurofibromin for NF1. Graphical Abstract O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=126 SRC="FIGDIR/small/734312v1_ufig1.gif" ALT="Figure 1"> View larger version (27K): org.highwire.dtl.DTLVardef@34b7eborg.highwire.dtl.DTLVardef@116c11aorg.highwire.dtl.DTLVardef@dfb251org.highwire.dtl.DTLVardef@341cb5_HPS_FORMAT_FIGEXP M_FIG C_FIG

genetics↗

bioRxiv2026

CWF19L2 couples pre-mRNA alternative splicing with the maternal-to-zygotic transition to safeguard female fertility

The maternal-to-zygotic transition (MZT) requires precise spatiotemporal execution of pre-mRNA alternative splicing (AS), yet the core splicing machinery driving this developmental reprogramming remains incompletely understood. Here, we identify the CWF19-like protein 2 (CWF19L2) as an indispensable pre-mRNA AS regulator that safeguards against oocyte and early embryo competence defects (OECD). While murine germline-specific depletion of Cwf19l2 spares morphological folliculogenesis, oocyte maturation, or fertilization, it induces complete female sterility characterized by profound developmental arrest at the 2-cell stage. Mechanistically, maternal deficiency of CWF19L2 localized to nuclear speckles disrupts transcription-splicing-translation coupling, collapsing AS homeostasis during maternal reserves and zygotic genome activation. We further demonstrate that CWF19L2 orchestrates the pre-mRNA splicing network through directly binding to target transcripts and indirectly modulating via interacting with the core spliceosomal factor PRPF8. Importantly, exogenous Cwf19l2 mRNA partially rescues the embryonic arrest. Together, our findings establish CWF19L2 as an indispensable AS engine during the MZT, providing a mechanistic foundation for OECD and a novel molecular etiology for female infertility.

Cell Biology↗

bioRxiv2026

Ciz1 safeguards Drosophila wing development by suppressing oxidative stress

Cell proliferation and fate specification are fundamental processes that ensure the generation of organs with proper size and patterning. Oxidative stress caused by accumulation of reactive oxygen species (ROS) can lead to cell cycle arrest, senescence, cell death and cell fate misspecification, thereby impairing normal development and contributing to many pathological processes. In this study, we identify Drosophila Ciz1 as a critical factor that safeguards epithelial homeostasis and development by preventing oxidative stress. Knockdown of Ciz1 in the Drosophila wing imaginal disc, an epithelial tissue that serves as the larval precursor of the adult wing, results in a small wing phenotype accompanied by thickened and ectopic veins. We further demonstrate that reduced Ciz1 expression leads to accumulation of donut-shaped mitochondria and elevated ROS levels. The increased oxidative stress subsequently suppresses proliferation via activation of JNK and promotes excessive vein formation by upregulating Rhomboid, a positive regulator of EGFR signaling. Interestingly, although Ciz1 is a zinc finger protein that predominantly localizes to the nucleus, neither its zinc finger motifs nor its nuclear localization is required for suppression of oxidative stress. Instead, the prion-like domain in its N-terminal part is essential for this activity. Our work identifies Ciz1 as an important factor in preventing oxidative stress and maintaining epithelial homeostasis.

Developmental Biology↗

bioRxiv2026

Metabolically engineered oilcane reshapes rhizosphere microbial guilds while preserving broad functional capacity

Metabolic engineering of crops can redirect host carbon flux, but its consequences for microbiomes remain unclear. Here, we show that engineering oilcane for triacylglycerol (TAG) accumulation reshapes rhizosphere microbial guilds across greenhouse and field environments while preserving functional capacity. Using 36 rhizosphere metagenomes from wild-type sugarcane and engineered oilcane accessions, we reconstructed metagenome-assembled genomes and linked community turnover with shifts in functional potential. Oilcane rhizospheres exhibited taxonomic restructuring relative to wild-type plants, driven primarily by turnover rather than nestedness and marked by genotype-dependent replacement of microbial guilds. These patterns were strongest in accession 1566 and amplified under field conditions. Despite these compositional shifts, broad patterns of functional potential remained similarly distributed, whereas pathway-level differences were evident in energy production and conversion, lipid transport and metabolism, secondary metabolite biosynthesis, transport and catabolism, and signal transduction. These findings extend evaluation of engineered crops beyond host traits alone to include microbiome-scale responses.

microbiology↗

bioRxiv2026

Epigallocatechin gallate and thermal cycling-stimulation synergistically promote apoptosis in A549 cells via endoplasmic reticulum stress-induced calcium ion dysregulation and oxidative stress

Non-small cell lung cancer (NSCLC), as the predominant subtype of lung cancer, presents a considerable clinical challenge due to its high rates of recurrence and the significant adverse effects associated with conventional therapeutic modalities. In response to these challenges, this study explored the new combined anticancer effects of epigallocatechin gallate (EGCG) together with thermal cycling-stimulation (TCS). The findings demonstrated that the combination of EGCG and TCS synergistically decreased the viability of A549 and NCI-H460 NSCLC cells, while exhibiting minimal cytotoxic effects on IMR-90 normal lung fibroblasts. Further investigation revealed that EGCG mitigated the TCS-induced upregulation of heat shock proteins HSP70 and HSP105 and concurrently diminished the expression levels of proteasome subunits. This combined effect disrupted proteostasis, resulting in pronounced endoplasmic reticulum (ER) stress. Subsequently, a positive feedback mechanism was established between inositol 1,4,5-trisphosphate receptor (IP3R)-mediated ER Ca2+ release and excessive reactive oxygen species (ROS) production, ultimately leading the cells to undergo mitochondrial apoptosis. This combined treatment reduces the necessary dosage of EGCG, thereby overcoming limitations related to its poor bioavailability and systemic toxicity, while also preventing the development of thermotolerance induced by TCS. Consequently, this method offers a new and potentially practical therapeutic strategy for treating NSCLC.

Cancer Biology↗

bioRxiv2026

Fibulin-2 transduces a matrix-to-metabolism signal in kidney fibrosis

Fibrotic extracellular matrix (ECM) is not merely a structural scaffold but an instructive signaling interface that shapes epithelial cell state. However, the molecular cues by which matrix remodeling controls tubular metabolism during kidney fibrosis remain poorly defined. Here, we identify Fibulin-2 (FBLN2) as a fibroblast-derived matrix cue that transduces fibrotic ECM remodeling into tubular mitochondrial metabolic reprogramming. Using fibroblast-selective deletion of Smoothened (Smo) across distinct fibroblast subpopulations, we found that loss of fibroblast Smo preserved kidney function and attenuated fibrosis in mouse models of chronic kidney injury. Multi-omics profiling revealed coordinated remodeling of the fibrotic matrisome, highlighted by suppression of FBLN2, an ECM glycoprotein genetically linked to kidney function in humans. Mechanistically, FBLN2 engaged EGFR in tubular epithelial cells and activated EGFR-AKT signaling in a non-canonical ligand-like manner. This signaling axis suppressed acetyl-CoA acetyltransferase 1 (ACAT1), a mitochondrial regulator of fatty acid oxidation and amino acid metabolism. Disruption of fibroblast Smo-FBLN2 signaling restored ACAT1-dependent oxidative metabolism and reduced tubular fibrotic activation. Spatial lipidomics revealed compartment-specific lipid remodeling associated with altered mitochondrial fatty acid metabolism, including acylcarnitine and phospholipid changes linked to reduced fibrotic injury. Together, these findings define a Fibulin-2-EGFR-ACAT1 matrix-to-metabolism signaling axis that couples fibrotic ECM remodeling to tubular mitochondrial metabolism during kidney fibrosis.

cell biology↗

bioRxiv2026

HoloCell: A Generative Foundation Model for Holistic Cellular Modeling

Single-cell multi-omics technologies have recently advanced to enable the profiling of epigenomic, transcriptomic, and proteomic layers within individual cells, offering new opportunities to characterize cellular states as integrated biological systems. However, developing a unified framework that can seamlessly integrate diverse omics modalities and remain robust to heterogeneous modality missingness remains challenging. Existing methods are often designed for specific modalities or modality pairs, relying on dataset-specific training or paired measurements. Here we present HoloCell, to our knowledge the first generative foundation model for joint representation learning and generative modeling across all three major single-cell omics modalities, i.e., epigenomics, transcriptomics, and proteomics. HoloCell contains over 860 million parameters and is pretrained on the Human-Multi-Omics-Corpus, which comprises approximately 468 million single-cell profiles across these three omics layers, corresponding to over 425 billion tokens. HoloCell introduces a a simple yet biologically motivated hierarchical tokenization strategy that encodes cis-regulatory elements, genes, and proteins as structured tokens within a shared modeling framework. We evaluated HoloCell across single-omics representation learning, paired multi-omics integration, unpaired multi-omics alignment, and cross-modal generation via iterative diffusion and remasking, demonstrating its superior performance and flexibility across diverse omics tasks. From a representation perspective, HoloCell provides a unified digital mapping of cellular states across multiple omics layers, capturing cell heterogeneity as an integrated system. From a generation perspective, its iterative diffusion and remasking frame-work permits flexible generation orders beyond fixed left-to-right causality, enabling in silico simulation of multi-omics information flow. Together, these capabilities position HoloCell as a versatile foundation model toward the emerging concept of a virtual cell, offering both systematic characterization and generative simulation of cellular systems within a unified framework.

bioinformatics↗