Biology subjects

Cai, J.

Publications and source records attributed to Cai, J..

At least 19 recordsLinked to original sources

bioRxiv2026

Uncoupling of STING Accumulation and Activation upon Chromosome-Derived Extrachromosomal Circular DNA Formation

The innate immune system maintains cellular homeostasis by recognizing cytosolic nucleic acids. While the cGAS-STING pathway is well characterized in detecting non-self-DNA, its response to self-DNA originating from the host nucleus remains unclear. Extrachromosomal circular DNA (eccDNA), a form of self-DNA often enriched in cancer, has been implicated in immune modulation, yet its ability to activate cGAS-STING remains unresolved. Here, we developed a Cre-loxP-based eccDNA reporter system, where Cre-mediated eccDNA formation triggers expression of the fluorescent protein mCitrine, enabling time-resolved, single-cell analysis of chromosome-derived eccDNA dynamics and STING activation. Using an mRuby3-STING reporter, we found that low-dose plasmid transfection induced STING activation, as evidenced by modest STING accumulation and downstream cytokine induction. Cells carrying chromosome-derived cytosolic eccDNA also showed a modest, cGAS-dependent accumulation of STING. However, this accumulation was uncoupled from detectable STING-Ser366 phosphorylation or induction of downstream cytokines. These findings indicate that chromosome-derived eccDNA does not drive detectable STING signaling under the conditions tested, and that STING accumulation and its activation are mechanistically separable steps. This work provides a foundation for understanding the immunological roles of eccDNA in genome instability and cancer.

cell biology↗

bioRxiv2026

CAI-Vax: Computational Design of Personalized mRNA Cancer Vaccine Constructs After Antigen Selection

CAI-Vax (Cancer Antigen Individualized Vaccine Design) arranges selected antigens and generates synonymous coding sequences for personalized mRNA vaccine constructs. We evaluated 357 vaccine-candidate peptides from 31 patients across four published cohorts. At the primary cumulative MHCflurry percentile-rank threshold of [≤]2%, HLA-aware search reduced predicted junction-spanning peptide-allele binder occurrences by 3.49 per junction relative to amino-acid-pattern optimization (paired Hodges-Lehmann estimate; nominal 95% interval 2.75-4.48; Holm-adjusted P = 7.45 x 10-). All 31 primary paired differences and the estimated effects in all 16 threshold-by-normalization settings favored HLA-aware search within the MHCflurry model used for optimization. Among 25 patients with an exact solution for the additive search objective, single-start greedy search found the minimum in 5 patients and the best individual heuristic in 16. Selecting each patients best result across all tested heuristics increased this to 19. MHCflurry and NetMHCpan EL identified identical tied best-plan sets in 6 of 31 patients after rescoring six frozen plans per patient. In a retrospective internal analysis, repeat-aware synonymous encoding reduced maximum 10-mer multiplicity in every patient, with mean codon-use-preference losses of 0.0041-0.0055 across cohorts; however, it performed worse than matched random encoding for 6-mers. These findings demonstrate that HLA-guided construct optimization can reduce predicted junctional binding within the optimization model, although the preferred antigen arrangement depends on the prediction model used. Experimental studies are needed to determine the biological and manufacturing relevance of these computational differences.

bioinformatics↗

bioRxiv2026

Venetoclax Resistance Creates Canonical Pyroptosis Vulnerability in AML

Venetoclax has transformed acute myeloid leukemia (AML) therapy, yet resistance is almost inevitable. Its diverse, often difficult-to-target drivers frequently converge on monocytic differentiation, a state that survives on MCL1 rather than BCL2. Here, we show that the very cells that escape BCL2 inhibition retain the canonical pyroptosis machinery of normal monocytes. Across 15 bulk and 4 single-cell cohorts, the canonical pyroptosis program was enriched in monocytic AML and monocytic leukemia stem cells, correlated with venetoclax resistance, and associated with inferior survival specifically after venetoclax-based therapy. In 24 cell lines and 142 patient specimens, both DPP8/9 inhibition by Val-boroPro and NLRP3 activation by lipopolysaccharide (LPS) plus nigericin preferentially killed monocytic AML cells through GSDMD-mediated pyroptosis, largely sparing non-monocytic AML and lymphoid leukemias. Venetoclax preferentially depleted primitive AML blasts but enriched a monocytic residuum vulnerable to canonical pyroptosis; combining venetoclax with Val-boroPro achieved deeper ex vivo clearance than either agent alone. In venetoclax-resistant xenografts, Val-boroPro suppressed leukemia progression, and adding venetoclax further reduced leukemia burden and prolonged survival beyond either monotherapy. Thus, venetoclax resistance itself exposes an alternative, druggable cell-death program that can be therapeutically exploited to overcome this resistance.

cancer biology↗

bioRxiv2026

Metagenomic insights into dose-dependent alterations in the tobacco rhizosphere microbiome by pyrroloquinoline quinone

Pyrroloquinoline quinone (PQQ) is known to promote plant growth, yet how the rhizosphere microbiome varies with PQQ concentration remains poorly understood. We applied metagenomic sequencing to characterize the taxonomic composition and functional potential of the tobacco rhizosphere microbiome across PQQ treatments at 0.5, 1 and 2 mol/L plus an untreated control. The rhizosphere microbiome was dominated by Pseudomonadota and Actinomycetota and differed among treatments in a concentration-associated, non-monotonic manner. The most pronounced differences were observed at 1 mol/L, where alpha diversity showed dimension-specific variation and beta-diversity analysis resolved the most distinct community profile. Taxonomic and functional profiles varied concurrently, with Streptomyces, Sphingomonas, Rhodanobacter and Lysobacter, together with functions related to biosynthesis, energy metabolism, colonization and environmental sensing, showing higher relative abundance at this concentration. No broad-spectrum antimicrobial effect was detected at the community level, and antibiotic resistance gene abundance showed no significant differences among treatments after correction for multiple testing. These findings provide a basis for further evaluation of PQQ in modulating the rhizosphere microbiome in continuous cropping systems.

microbiology↗

bioRxiv2026

A stromal metabolic program suppresses NK-cell immunity to drive tumor progression in HER2-low breast cancer

Cancer-associated fibroblasts (CAFs) are major regulators of the tumor microenvironment, yet how distinct CAF states suppress innate immunity in HER2-low breast cancer remains poorly understood. Here, we identify an S100A4-enriched CAF population that expands during HER2-low breast tumor progression and establishes a metabolically immunosuppressive niche. Spatial transcriptomics and multiplex imaging of human HER2-low tumors reveal progressive CAF accumulation and an inverse spatial association between S100A4-enriched CAFs and immune infiltration, including natural killer (NK) cells. Using an immunocompetent HER2-low mammary tumor model, we show that S100A4-enriched CAFs promote tumor initiation and progression while suppressing NK-cell cytotoxicity, IFN-{gamma} production, perforin, and granzyme B. Fractionation of CAF-conditioned media and metabolic profiling identify a low-molecular-weight immunosuppressive program characterized by enhanced branched-chain amino acid catabolism and accumulation of branched-chain -keto acids (BCKAs). Mechanistically, BCKAs directly suppress NK-cell IFN-{gamma} production, whereas inhibition of the branched-chain aminotransferase BCAT1 reduces CAF-mediated NK-cell suppression and restores antitumor cytotoxicity. BCAT1 inhibition also suppresses HER2-low tumor growth in vivo, an effect attenuated by NK-cell depletion, establishing NK-cell restoration as a functional component of its antitumor activity. Together, these findings uncover a CAF-driven metabolic immune checkpoint in which S100A4-enriched CAFs exploit BCAT1-dependent BCKA production to suppress NK-cell surveillance and promote HER2-low breast tumor progression. Targeting stromal BCAT1 therefore represents a potential strategy to dismantle CAF-mediated immune suppression and restore innate antitumor immunity.

cancer biology↗

bioRxiv2026

Method Choice, Not Biology, Determines In Silico Perturbation Results: A Systematic Evaluation of Eight Methods Across Four Datasets

Most in silico perturbation methods for single-cell transcriptomics have been validated only on individual datasets, leaving their reliability and generalizability unknown. Through systematic cross-method, cross-dataset benchmarking of eight methods spanning six mathematical frameworks across four datasets, we find that six of eight methods--including widely used VAE-based and tensor decomposition approaches--fail to produce detectable transcription factor (TF)-to-pathway signals. Only CellOracle and DDIM consistently detected TF-to-glycolysis directional regulation. Cross-pathway analysis in PBMC monocytes revealed biologically coherent TF-pathway associations beyond glycolysis (SPI1[->]glycolysis 4.4x enrichment, FOS[->]AP-1 targets 4.4x), with SOX9 serving as a biological specificity control (no pathway enrichment). Method choice alone could reverse biological conclusions: DDIM and scTenifoldKnk rankings were significantly anti-correlated ({rho}=-0.811, p=0.027). CRISPRi Perturb-seq validation in K562 cells confirmed TF knockdown suppresses glycolysis gene expression (JUN {delta}=-1.72, CEBPB {delta}=-1.59, SPI1 {delta}=-1.57, FOS {delta}=-0.70), but CellOracle-predicted perturbation directions did not match experimental directions (40.9% agreement, not different from chance), revealing a fundamental gap between steady-state correlation and causal perturbation. Diagnostic analyses using VAE latent space profiling, correlation distribution comparison, and gene-gene graph analysis identified distinct failure modes in unsuccessful methods: VAE latent space competition (STAT3 signal-to-noise 0.44 vs. SPI1 4.25), correlation noise (TF-glycolysis |r|=0.038 indistinguishable from background |r|=0.047), and graph non-specificity (0.84x enrichment). A controlled ablation experiment showed that adding a GRN prior to DDIM did not improve target recall (delta=0 for all TFs), confirming that performance differences are multi-factorial. These findings establish preliminary guidance for method selection, including cross-pathway validation, direction-aware benchmarking, and minimum data requirements ([≥]500 cells, [≥]1,000 HVGs).

bioinformatics↗

bioRxiv2026

Nutrient flux governs osteogenic fate commitment through the SLC3A1-cystine Axis

Skeletal and mesenchymal cells have limited bone-forming potential due to their rarity, tendency for senescence and/or unstable osteogenic lineage commitment. We previously identified an osteopotent CXCR4 stem cell population with unclear mechanism for its differentiation potential. Here, we found that the cystine transporter SLC3A1 was selectively enriched in CXCR4+ stem cells, uncovering a role for amino acid transport in regulating osteogenic fate commitment. Enforced SLC3A1 expression reprogrammed mesenchymal cells toward a stable osteogenic state while suppressing adipogenic differentiation. Mechanistically, SLC3A1-mediated cystine flux established a glutathione-dependent metabolic program that preserved mitochondrial fitness and restrained stem cell senescence. SLC3A1 also stabilized IGF-1 through suppression of ZMYND8-mediated ubiquitination, uncovering ZMYND8 as a previously unrecognized E3 ligase regulating osteogenic commitment. Cystine supplementation phenocopied the effects of SLC3A1 activation, promoting osteogenic differentiation and skeletal regeneration without genetic manipulation. In vivo, cystine administration accelerated bone regeneration and attenuated ovariectomy-induced bone loss, while analyses of a human osteoporosis cohort, osteoporotic specimens, and single-cell transcriptomic datasets revealed coordinated suppression of the SLC3A1-cystine-IGF-1 pathway in osteoporotic mesenchymal cells. Taken together, these findings establish SLC3A1-mediated cystine transport as a programmable metabolic determinant of osteogenic fate commitment and identify cystine metabolism as a therapeutically targetable axis for skeletal regeneration and osteoporosis.

cell biology↗

bioRxiv2026

A Synthetic Microbial Therapy Rewires Antitumor Immunity Across Multiple Cancer Types

Microbial immunotherapies show promise against cancer, yet broad efficacy and mechanistic insight remain elusive. Here, we introduce SPIKE 1.0 (S1.0), a metabolically engineered bacterium that converts tryptophan into immunomodulatory hydroxyindoles via tryptophan monooxygenase to remodel the tumor microenvironment (TME). A single systemic dose of S1.0 elicited potent, durable antitumor responses across multiple murine models, including humanized mice, with minimal toxicity. S1.0 enhanced inflammatory signaling, activated innate and adaptive immunity, and promoted T cell persistence, memory, and resistance to exhaustion. It outperformed checkpoint inhibitors and synergized with chemotherapy. Multi-omics profiling revealed that S1.0 rewired amino acid metabolism in tumor-infiltrating immune cells and disrupted immunosuppressive networks. These results establish S1.0 as a scalable, cost-effective microbial immunotherapy with broad translational potential for solid tumors.

immunology↗

bioRxiv2026

Hypothalamic Representation of Aggressiveness across Mouse Strains

Aggression is an innate behavior conserved across species, serving as a critical means to compete for food, mating opportunities, and other essential resources. A central question in aggression research is the extent to which inter-individual variability in aggression is shaped by genetic factors. Here, we examine aggressive behaviors in naive male mice across seven genetically defined strains and find large cross-strain differences. We find a tight correlation between aggressiveness and anxiety levels across strains, but not within the same strain, suggesting strong genetic control of both traits. Pharmacologically elevating anxiety in high-aggression strains reduces aggression, revealing a causal relationship between these behaviors. We further demonstrate that differences in the synaptic and cellular properties of neurons in the ventrolateral ventromedial hypothalamus (VMHvl) largely account for cross-strain variability in male aggression, and that chemogenetically increasing VMHvl excitability enhances attack behavior in a low-aggression strain. Together, these findings reveal the neuronal implementation of the genetic control of innate aggression level.

neuroscience↗

bioRxiv2026

Genos-m: a foundation model for human-associated microbial genomes

Human-associated microbial genomes encode extensive strain-level diversity and niche-specific gene repertoires that are critical to host health. However, these complex sequence features remain difficult to capture using general-purpose DNA foundation models, highlighting the need for dedicated representation learning tailored to the human microbiome. Here, we introduce Genos-m, an open-source foundation model for human-associated microbial genome representation. Genos-m was pretrained on approximately 1.2 trillion nucleotide tokens from a curated microbial genome corpus, including human-associated prokaryotic isolates, high-quality metagenome-assembled genomes (MAGs) and bacteriophages, supplemented with GTDB species-level representative genomes to broaden prokaryotic taxonomic breadth. The model uses a sparsely activated Mixture-of-Experts (MoE) Transformer architecture, with 4.7 billion total parameters, approximately 330 million activated parameters per forward pass and a maximum context length of one million base pairs. We evaluated frozen Genos-m representations across short-sequence and gene-level tasks, biosynthetic gene cluster (BGC)-based regional sequence tasks, whole-genome strain phenotype prediction, and zero-shot transfer on prokaryote-related RNAfitness assays. Across these benchmarks, Genos-m consistently ranked among the leading comparison models, with the best performance in five of eight gene-fitness regression tasks and in BGC type classification. Using sparse autoencoders, we identified sparse features in Genos-m hidden activations that aligned with annotated ORFs, intergenic regions, and tRNA and rRNA loci. In downstream applications, Genos-m-derived genome-informed species representations in-corporated into a human microbiome self-supervised learning model improved colorectal cancer (CRC)-control classification over conventional species-abundance random forest models. Genos-m also generated stable sample-level embeddings from as few as 10,000 metagenomic reads, retaining gut microbial community structure that distinguished geographic origin and aligned with enterotypes defined from full-depth taxonomic profiles. Together, these results support Genos-m as a reusable representation model for microbial genomes and metagenomes, with conclusions bounded by the reported datasets, task definitions and evaluation protocols. Genos-m model weights, inference code, and usage documentation are publicly available on GitHub (https://github.com/BGI-HangzhouAI/Genos-m) and Hugging-Face (https://huggingface.co/BGI-HangzhouAI/Genos-m).

bioinformatics↗

bioRxiv2026

Purinergic signaling promotes gliomagenesis through nuclear calcium transients

Intracellular Ca2+ transients drive key developmental and physiological processes, yet their role in oncogenesis remains incompletely understood. In glioblastoma (GBM), an aggressive brain malignancy, tumor cellular networks exhibit self-sustaining Ca2+ transients that promote tumor growth through unclear mechanisms. Using patient-derived GBM models, we show that these transients depend primarily on intracellular Ca2+ stores and extend to the nucleus to drive tumorigenesis. A neuromodulator screen identified extracellular purines ATP and ADP as potent inducers of both nuclear and cytosolic Ca2+ transients via activation of metabotropic purinergic P2RY1 receptors, whose knockdown attenuates tumorigenicity in vitro and in vivo. Mechanistically, Ca2+ transients promote tumorigenesis via the nuclear Ca2+/calmodulin-dependent kinase CAMK4, which regulates transcriptional and epigenetic programs, as well as ribosomal DNA transcription. From the therapeutic perspective, pharmacologic P2RY1 inhibition suppresses tumor growth in vitro and in vivo. Collectively, these findings reveal a pharmacologically targetable oncogenic mechanism in GBM and possibly other malignancies.

cancer biology↗

bioRxiv2026

3D-PAULM: Integrated Photoacoustic Tomography and Ultrasound Localization Microscopy for Multiparametric Brain and Tumor Imaging

Understanding processes such as blood-brain barrier (BBB) disruption and tumor progression can greatly benefit from simultaneous molecular, functional, and hemodynamic imaging in deep tissue, yet few existing imaging modalities can provide all three in a single system. Here, we present an integrated imaging platform that combines 3D photoacoustic tomography with ultrasound localization microscopy (3D-PAULM) to enable intrinsically co-registered, multiparametric imaging. 3D-PAULM unifies multispectral photoacoustic molecular imaging, ultrasound B-mode imaging, microbubble-enhanced power Doppler, and ultrasound localization microscopy, and concurrently measures blood oxygenation, blood perfusion, microvascular flow dynamics, and molecular probes from near-infrared dyes and photoswitchable phytochromes. We apply 3D-PAULM to quantify BBB leakage in focal ischemia and systemic inflammation, and to perform high-sensitivity molecular imaging of solid tumors alongside functional mapping of tumor hypoxia and super-resolved vascular remodeling. Together, these results establish 3D-PAULM as a versatile platform for integrated functional and molecular imaging in deep tissue.

bioengineering↗

bioRxiv2026

Characterization of the SKC mouse strain as a potential model for keratoconus

BackgroundKeratoconus (KC) is a multifactorial disorder with unclear etiology, characterized by localized thinning and a cone-like protrusion of the cornea. The complex etiology of KC exacerbates the lack of an animal model. Previous studies by Tachibana et al. (2002) described an inbred mouse strain (SKC) with a spontaneous, androgen-dependent, cone-like corneal morphology. This study aimed to investigate the corneal phenotypes of SKC mice through an in-depth ophthalmic examination. MethodsMice (n=53) were examined via slit lamp biomicroscopy with fluorescein staining. Spectral-domain optical coherence tomography (SD-OCT) enabled central corneal thickness (CCT) measurement in selected mice (n=26 eyes), and OCT-based pachymetry mapping (n=16 eyes). In vivo corneal confocal microscopy was conducted on eyes to assess cellular morphology (n= 9 eyes). Eyes were collected for histology analysis (n=22). ResultsLesions and epithelial breaks were present in [~]95% of eyes (n=101). Neovascularization, perforation, scarring, and hydrops were seen primarily in males. An opaque, unilateral cone-like morphology was exclusive to males (n=11). Male and female corneas showed no significant difference in CCT, though pachymetry mapping revealed regional thinning patterns in both sexes. Loosened epithelial tight junctions, stromal fibrosis, vascularization, and inflammation of variable severity were identified in both sexes. ConclusionThis study identified previously unreported corneal phenotypes in SKC mice through ophthalmic examination. Unlike previous studies, gross and histological abnormalities were observed in female SKC mice. Our findings suggest a lower penetrance of the cone-like phenotype ([~]20%) than previously reported ([~]33%) and support that the conical phenotype in male mice may be secondary to keratitis.

pathology↗

bioRxiv2026

Rewiring Oncogenic Transcriptional Complexes with Domain-ALTeration Chimeras (DALTACs) in Prostate Cancer

Transcriptional addiction to the androgen receptor (AR) underlies metastatic castration-resistant prostate cancer (mCRPC), where AR maintains oncogenic programs through domain-specific interactions with the acetyltransferases p300/CBP. Here, we describe a distinct therapeutic modality, Domain-ALTerAtion Chimeras (DALTACs), that rewire endogenous protein complexes by enforcing non-native domain-domain interactions. Our first-in-class AR-p300/CBP DALTAC-1 induces synthetic proximity between the AR ligand-binding domain and p300/CBP-bromodomain, misconfiguring their native interface and locking the complex in a non-productive state. DALTAC-1 triggers a super-inhibitory effect, suppressing AR-driven transcription and proliferation more potently than combined AR and p300/CBP inhibition. Mechanistically, DALTAC-1 reprograms p300/CBP substrate engagement, abolishing the histone mark H2B N-terminal acetylation (H2BNTac) while inducing neomorphic acetylation of AR and coactivators, ultimately collapsing the AR neo-enhanceosome. Chromatin profiling revealed redistribution of AR and p300/CBP toward palindromic AREs, accompanied by disruption of cofactor and transcriptional machinery recruitment at oncogenic AR/ERG neo-enhancers. DALTAC-1 exhibits exquisite lineage selectivity, displaying potent activity in AR-p300/CBP dual-dependent cells while sparing AR-negative cells. In multiple in vivo models, DALTAC-1 induces deep and durable tumor regressions. Together, these findings establish DALTACs as a broadly applicable strategy to rewire disease-defining complexes through domain-topology alteration, expanding induced-proximity therapeutics, while DALTAC-1's precise lineage-selectivity supports its translational potential in AR-driven prostate cancer.

cancer biology↗

bioRxiv2026

Deep mapping of structural perturbations to energetics enables precise TCR design

Affinity optimization and cross-reactivity profiling are pivotal for T-cell receptor (TCR) engineering and immunotherapy, yet remain hindered by the vast diversity of the TCR repertoire and limited structural insights. Here, we present mpTCRai, a deep learning framework that captures the sequential characteristics of antigen presentation and recognition with high precision. Leveraging these precise structural predictions, we established a contact hotspot-based scoring mechanism that explicitly maps structural perturbations to energetic changes, yielding a strong correlation with experimental affinity (r = -0.88). In application, the model identified critical molecular switches, such as specific G-to-A substitutions, and revealed their dependence on structural context. Furthermore, it effectively distinguished functional variants from dangerous cross-reactive mutations like Y5W, thereby mitigating the risk of off-target toxicity. Guided by these insights, we computationally designed four novel A6-TCR variants, demonstrating a rational strategy for candidate selection in adult T-cell leukemia. This work establishes a unified platform integrating structural and energetic constraints to advance precise, safe TCR therapeutic design.

immunology↗

bioRxiv2026

Bacterial internalins exploit E-cadherin to promote head-neck tumor metastasis and drug resistance

Head and neck squamous cell carcinoma (HNSCC) is an aggressive malignancy characterized by local invasion, lymph node metastasis, and therapeutic resistance. Chronic periodontal disease has been linked to HNSCC progression, yet the responsible pathogens and underlying molecular mechanisms remain unclear. Here, we show that the keystone periodontal pathogen Porphyromonas gingivalis promotes HNSCC metastasis and chemoresistance through two internalin proteins that are secreted via the type IX secretion system (T9SS). These internalin proteins specifically bind the EC1 domain of E-cadherin through their curved solenoid-like leucine-rich repeats (LRRs), facilitating bacterial invasion and inducing epithelial-to-mesenchymal transition (EMT). Mechanistically, internalin-E-cadherin engagement drives {beta}-catenin nuclear translocation and activates p38 and JNK1/2 MAP kinase signaling pathways, enhancing tumor cell migration, metastatic dissemination, and resistance to cisplatin-induced apoptosis. Tissue microarrays detect internalin antigens in HNSCC specimens, supporting their in vivo relevance. Together, these findings establish a direct mechanistic link between an oral pathogen and HNSCC progression and extend the paradigm of internalin-E-cadherin interactions from microbial pathogenesis to cancer biology.

microbiology↗

bioRxiv2026

VeloTrace Reconciles Divergent Velocity and Trajectory in Single-cell Transcriptomics with Deep Neural ODE

Cellular identity and fate transitions are governed by continuous molecular processes that form dynamic trajectories within a high-dimensional transcriptomic landscape. Existing methods attempt to model these dynamics from two complementary perspectives: trajectory inference and velocity modeling. Ideally, velocity and trajectory are dual aspects of transcriptomic dynamics where velocity is tangent to trajectory everywhere. This inherent connection between velocity and trajectory is currently absent in transcriptomic analysis. Splicing velocity are precision-limited to inadequately-sequenced genes, while trajectory inference prioritizes the modeling of global trends while omitting local dynamics. This divergence breaks the geometric continuity between local velocities and global trajectories, hindering the reliable interpretation of developmental dynamics. To reconcile trajectory inference and RNA velocity, we introduce VeloTrace, a framework that unifies them through Neural Ordinary Differential Equations (NeuralODEs). VeloTrace learns a continuous-time velocity field whose integral curves constitute the trajectory itself, while ensuring that velocities are tangent to integral paths everywhere. Leveraging a splicing quality score, VeloTrace incorporates high-quality splicing velocity as partial supervision for velocity orientation and grounding. During optimization, VeloTrace incorporates a Monte Carlo multi-time-frame supervision strategy to ensure coherence between local and global trajectorys and suppress sequencing-induced stochastic diffusion. Through refining the velocity field and cell-specific parameters for pseudo-time, expression, and velocity, VeloTrace reconstructs a smooth, local-and-global-coherent velocity-vector-guided flow in the transcriptomic latent space. This strategy ensures a complementary integration of velocity and trajectory, imputing the transcriptional kinetics for genes of insufficient strength, whose kinetics cannot be accurately portrayed by splicing velocity. In simulation benchmarks, VeloTrace captured the transcriptional dynamics of all expressed genes, even those with inadequate sequencing coverage, producing velocity directions that were most consistent with the true direction and every-where tangential across the entire process, outperforming state-of-the-art methods, including scVelo, UniTVelo, VeloVI and scTour. VeloTrace uniquely reconciles RNA velocity and trajectory inference, creating a velocity field where each cell can infer past and future transitions from its current state. Moreover, VeloTrace extends reliable velocity estimation to a broader set of genes. When applied to mouse neural stem cell differentiation data, it successfully recovers dynamics of driver genes for two developmental lineages, including those with low expression, shedding light on their regulatory roles during differentiation. This unified framework lays the foundation for more accurate modeling of gene regulation and cell fate decisions in complex biological systems.

bioinformatics↗

bioRxiv2026

Antibiotic Exposure Through Human Milk Influences the Infant Gut Microbiome

Infant antibiotic treatment is associated with increased risk of developing non-communicable diseases, potentially through disruption of the gut microbiome. However, the impact of indirect antibiotic exposure via human milk remains largely unexplored. Here, we investigate a cohort (n=80) of antibiotic-treated breastfeeding mother-infant dyads and untreated matching controls using integrative multi-omics analyses of fecal, milk, and skin samples (n=1,455). Maternal antibiotic treatment was associated with different infant fecal microbiome and metabolome profiles, including lower abundance of Bacteroides, Lactobacillus, and Bifidobacterium, and higher levels of antimicrobial resistance gene reads. Further, fecal metabolic alterations associated with indirect antibiotic exposure were exacerbated by formula milk supplementation. In a subset of infants (n=61), indirect exposure was associated with higher body mass index (BMI). These findings suggest that maternal antibiotic treatment during lactation may influence the early-life infant gut microbiome with potential long-term implications.

microbiology↗