Search bioRxiv⌕ Search

Biology subjects

Sinha, S.

Publications and source records attributed to Sinha, S..

At least 37 records · Page 2Linked to original sources
bioRxiv2026

An Endocytic Checkpoint Controls Macrophage PD-1 Function and Immunotherapy Fate

Responses to PD1 blockade span durable tumor control to hyperprogressive disease (HPD), yet innate immune mechanisms governing these extremes remain undefined. Here we integrate a macrophage systems atlas (>12,500 transcriptomes) with single-cell profiles from >1,000 anti-PD1 treated patients, to identify CCDC88A (GIV) as a macrophage-intrinsic determinant of durable response versus HPD. GIV loss increases PD1 surface retention, suppresses phagocytosis, and accelerates tumor growth across murine models, human macrophages, and patient-derived organoids. Myeloid-specific GIV deletion converts PD1 blockade from tumor-restraining to tumor-accelerating by reprogramming macrophages toward HPD-like states. Mechanistically, GIV engages a conserved TIR-like [TILL] motif within the PD1 cytoplasmic tail to drive dynamin-dependent endocytosis, coupling innate immune signaling logic to checkpoint receptor trafficking. Pharmacologic disruption of this axis phenocopies GIV loss, revealing an endocytic vulnerability that undermines checkpoint efficacy and triggers accelerated growth at relapse. These findings define PD1 routing, rather than ligand-binding, as a macrophage-encoded checkpoint governing antitumor immunity. O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=200 SRC="FIGDIR/small/718292v1_ufig1.gif" ALT="Figure 1"> View larger version (65K): org.highwire.dtl.DTLVardef@a8f83aorg.highwire.dtl.DTLVardef@140dc7dorg.highwire.dtl.DTLVardef@14bc366org.highwire.dtl.DTLVardef@1096b8a_HPS_FORMAT_FIGEXP M_FIG GRAPHIC ABSTRACT C_FIG In briefMullick et al. show that PD1 trafficking in tumor associated macrophages (TAMs), controlled by the endocytic adaptor GIV/Girdin, determines the durability of immunotherapy. Disrupting this pathway traps PD1 at the surface, converts checkpoint blockade into hyperprogressive disease, and abolishes therapeutic benefit, revealing PD1 routing as a macrophage-encoded checkpoint of immunotherapy outcome. HighlightsO_LIGIV defines macrophage states that stratify durable response vs hyperprogressive disease C_LIO_LIGIV limits surface checkpoint accumulation by facilitating PD1 endocytosis C_LIO_LILoss of GIV traps PD1 at the membrane and drives hyperprogressive disease C_LIO_LIEndocytic blockade phenocopies GIV loss and abolishes checkpoint efficacy C_LIO_LIPsychotropic drugs disrupt PD1 trafficking and accelerate tumor growth C_LI

cancer biology↗
bioRxiv2026

Benchmark of biomarker identification and prognostic modeling methods on diverse censored data

The practices of identifying biomarkers and developing prognostic models using genomic data has become increasingly prevalent. Such data often features characteristics that make these practices difficult, namely high dimensionality, correlations between predictors, and sparsity. Many modern methods have been developed to address these problematic characteristics while performing feature selection and prognostic modeling, but a large-scale comparison of their performances in these tasks on diverse right-censored time to event data (aka survival time data) is much needed. We have compiled many existing methods, including some machine learning methods, several which have performed well in previous benchmarks, primarily for comparison in regards to variable selection capability, and secondarily for survival time prediction on many synthetic datasets with varying levels of sparsity, correlation between predictors, and signal strength of informative predictors. For illustration, we have also performed multiple analyses on a publicly available and widely used cancer cohort from The Cancer Genome Atlas using these methods. We evaluated the methods through extensive simulation studies in terms of the false discovery rate, F1-score, concordance index, Brier score, root mean square error, and computation time. Of the methods compared, CoxBoost and the Adaptive LASSO performed well in all metrics, and the LASSO and elastic net excelled when evaluating concordance index and F1-score. The Benjamini-Hoschberg and q-value procedures showed volatile performances in controlling the false discovery rate. Some methods performances were greatly affected by differences in the data characteristics. With our extensive numerical study, we have identified the best performing methods for a plethora of data characteristics using informative metrics. This will help cancer researchers in choosing the best approach for their needs when working with genomic data.

bioinformatics↗
bioRxiv2026

A NOD2-Encoded Toggle Switch Resolves the Host-Microbe Battle Over Cyclic AMP Control

Pathogens hijack macrophages by triggering pathological cAMP surges that block phagolysosomal killing--a defect mirrored in phagocytes from refractory colitis. We identify a host-encoded, pathogen-specific surge-protector comprised of a three-protein toggle: The innate sensor NOD2 binds and masks an evolutionarily conserved motif in GIV that activates trimeric-GTPase Gi, enforcing a biphasic surge-to-plunge cAMP-program: early, NOD2*GIV assembly permits a brief, tolerogenic cAMP rise, whereas subsequent GIV*Gi engagement collapses cAMP to drive phagolysosomal fusion and microbial clearance. Structural, biochemical, and ultrastructural analyses reveal how molecular toggling imposes precise spatial and temporal control. Pharmacogenomic perturbations pinpoint cAMP-PKA hyperactivation as the defining lesion in GIV-deficient macrophages. Functional studies in primary macrophages and human gut organoid co-cultures show that toggling the NOD2*GIV*Gi-axis is necessary and sufficient to convert tolerant macrophages into microbicidal machines that preserve mucosal barrier integrity. These findings uncover a druggable cAMP-control pathway with therapeutic promise in colitis. GRAPHIC ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=200 SRC="FIGDIR/small/715116v1_ufig1.gif" ALT="Figure 1"> View larger version (70K): org.highwire.dtl.DTLVardef@53cd40org.highwire.dtl.DTLVardef@1c83c82org.highwire.dtl.DTLVardef@fcbc58org.highwire.dtl.DTLVardef@b14f8e_HPS_FORMAT_FIGEXP M_FIG C_FIG eTOC BlurbPathogens hijack macrophages by inducing cAMP surges that help them evade clearance. Anandachar et al. identify a host "toggle switch" in which NOD2 and G proteins compete for GIV, driving a rapid and robust surge-to-plunge transition in cAMP. This temporal switch limits tolerogenic signaling, restores microbial clearance and barrier integrity, and unveils a targetable host pathway in infection and IBD. HighlightsO_LIPathogens exploit cAMP surges in macrophages to block phagolysosomal killing of microbes C_LIO_LIGIV acts as a molecular "toggle" linking NOD2 sensing to Gi-mediated cAMP control C_LIO_LIStructural and mutagenesis studies reveal mutually exclusive binding of NOD2 and Gi to GIV C_LIO_LIPharmacogenomic perturbations pinpoint PKA, not EPAC, as the critical downstream effector C_LIO_LIOrganoid co-cultures show NOD2-GIV-PKA crosstalk safeguards microbial clearance and gut barrier integrity C_LI

immunology↗
bioRxiv2026

Gradient Multinozzle 3D Printing

Direct ink writing is compatible with an expansive materials palette. While enabling diverse applications, this materials versatility brings significant bottlenecks in ink formulation, often requiring the mixing, printing, and testing of dozens to hundreds of ink compositions over the course of a project. To accelerate ink-space exploration, we introduce gradient embedded multinozzle (GEM) printheads that combine the high-throughput parallelized printing of multinozzles with combinatorial ink mixing. These printheads allow simultaneous mixing of two-, three-, and four-input inks which are distributed to printer nozzles to create complex 3D structures with graded compositions of inks. Using a two-way GEM printhead, we vali-date cell compatibility by printing scaffolds containing various concentrations of fibroblasts and observing non-linear compaction behaviours. We next test a three-way GEM multinozzle to print ten compositions of di- and multi-functionalized poly(ethylene-glycol) diacrylate hydrogel tri-leaflet valves, optimizing for stiffness, swelling ratio, and toughness. Our GEM multinozzles are compatible with open-source printers and either pressure- or volume-driven extrusion systems and promise to accelerate iterative ink design and testing.

bioengineering↗
bioRxiv2026

cGAS inhibition delays TDP-43-driven ALS Pathogenesis

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disorder marked by motor neuron loss and cytoplasmic mislocalization of TAR DNA-binding protein 43 (TDP-43), a key regulator of RNA splicing. However, the upstream modulators of this process remain poorly defined. Here we identify cyclic GMP-AMP synthase (cGAS) as a central mediator of TDP-43 pathology and associated mis-splicing. cGAS expression was elevated in ALS patient brains and enriched across activated microglia. In human iPSC-derived microglia-motor neuron co-cultures, neuronal TDP-43 pathology triggered microglial cGAS activation, whereas pharmacological inhibition with a potent human cGAS inhibitor reduced phosphorylated TDP-43, restored lysosomal and phagocytic programs, normalized microglial reactivity, and reversed TDP-43-associated RNA splicing defects. In vivo, cGAS inhibition in TDP-43 Q331K mice reversed widespread RNA splicing abnormalities across neurons and oligodendrocyte lineage cells, attenuated neurodegenerative pathology, and preserved motor function. Together, these findings identify cGAS as a druggable upstream regulator linking innate immune signaling to TDP-43-dependent RNA mis-splicing and neurodegeneration, and establish cGAS inhibition as a promising therapeutic strategy for ALS.

neuroscience↗
bioRxiv2026

A Druggable G Protein Checkpoint in Cholesterol Efflux

Immunometabolic diseases such as obesity, fatty liver, and atherosclerosis arise when lipid-associated macrophages (LAMs) fail to resolve lipid overload via reverse cholesterol transport (RCT), the bodys sole pathway for lipid disposal. How RCT is restrained in disease remains unknown. Integrating systems modeling with human plaque transcriptomes, we identify LAM subpopulations associated with plaque progression and uncover CCDC88A (GIV) as a macrophage-intrinsic checkpoint on RCT. Myeloid-specific GIV deletion in mice reduces aortic plaque burden, mobilizes hepatic and adipose lipid stores and promotes fecal sterol disposal. Mechanistically, GIV sequesters the cholesterol transporter ABCA1 within endomembranes and activates Gi[bullet]{beta}{gamma} signaling to suppress cAMP/PKA-CREB-dependent efflux programs. Pharmacogenomic disruption of this checkpoint reactivates efflux programs. In human plaque-in-a-dish models targeting this pathway restored efflux where statins and {beta}-blockers failed, translating to an estimated [~]98% reduction in plaque-progression risk in outcome modeling. Thus, RCT-restoration represents a macrophage-intrinsic therapeutic paradigm for immunometabolic disease. GRAPHIC ABSTRACT O_FIG O_LINKSMALLFIG WIDTH=200 HEIGHT=199 SRC="FIGDIR/small/705962v2_ufig1.gif" ALT="Figure 1"> View larger version (64K): org.highwire.dtl.DTLVardef@50dd10org.highwire.dtl.DTLVardef@e16d9borg.highwire.dtl.DTLVardef@1e2eaf3org.highwire.dtl.DTLVardef@af29b9_HPS_FORMAT_FIGEXP M_FIG C_FIG eTOC blurbLipid-associated macrophages drive immunometabolic disease. Katkar et al. show that disabling a GIV-dependent G-protein brake restores cholesterol efflux, reverses plaque lipid accumulation, and establishes reverse cholesterol transport as a druggable therapeutic axis. HighlightsO_LIStatins slow but rarely reverse plaque burden, leaving residual risk driven by LAM dysfunction C_LIO_LIGIV (CCDC88A) non-canonically modulates Gi to suppress macrophage cholesterol efflux C_LIO_LIGIV loss or inhibition restores ABCA1 activity via transcriptional and post-translational control C_LIO_LIBlocking GIV[bullet]Gi checkpoint defats LAMs, regresses plaques, and relieves systemic lipid overload C_LIO_LISIdentifies a druggable node that redefines RCT restoration as a therapeutic paradigm in immunometabolic disease C_LI

immunology↗
bioRxiv2026

MANTIS: Analytics toolkit for spatial metabolomics with matching spatial transcriptomics data

MotivationJoint Spatial Metabolomics (SM) and Spatial Transcriptomics (ST) profiling is a powerful approach to fine-mapping of metabolic states associated with tissue function. Current computational tools for analysis of "SM+ST" data focus primarily on alignment and integration of the two modalities, with limited support for probing biological relationships between the two molecular layers. ResultsWe present MANTIS, a statistical framework for analyzing co-registered SM+ST profiles at single cell or spot resolution, along with spatial domain or cell type information, to discover metabolite spatial patterns and gene-metabolite relationships. It employs an autocorrelation-preserving permutation strategy to assess statistical significance, yielding calibrated inference under spatial dependence. It disentangles different sources of spatial patterns and correlations, viz., those arising from regional preferences, cell type associations, or other unknown factors. It introduces the use of spatial cross-correlation and spatial partial correlation statistics for quantifying gene-metabolite associations. Across data sets spanning different spatial technologies, tissues and species, MANTIS provides more specific and interpretable discoveries than existing methods through rigorous statistical testing and explicitly modeling confounding structure. To our knowledge, MANTIS is the first toolkit to unify spatial metabolomics, spatial transcriptomics, cell type information and spatial domains within a single framework that emphasizes spatial statistics, hypothesis testing and confounder correction. Availability and ImplementationFreely available on the web at https://github.com/yuhaotuo/MANTIS.

bioinformatics↗
bioRxiv2026

CELLWHISPER: INFERENCE OF DIRECT CELL-CELL COMMUNICATION FROM SPATIAL TRANSCRIPTOMICS

AbstractDirect, contact-dependent cell-cell communication shapes tissue physiology and disease. Spatial transcriptomics offers a unique opportunity to infer such communication mechanisms, but the few existing tools for such inference are prone to high error rates due to spatial structure in distribution of cell types and gene expression. We present CellWHISPER, a statistical framework for inferring contact-mediated cell-cell communication, including gap junctions and ligand-receptor mechanisms, from single-cell resolution spatial data. CellWHISPER achieves strict error control via a specialized hypothesis-testing procedure that accounts for the confounding effects of cell- type-specific expression and spatial organization. Its use of an exact closed-form significance score obviates expensive permutation testing and enables scalable analysis of large tissues an extensive signaling compendia. To uncover salient communication patterns, CellWHISPER also includes a latent variable model that distills recurrent patterns into maps of mutual preferences among cell types and signaling genes. Applied to mouse brain STEREO-seq and Xenium datasets, CellWHISPER identifies cell type-specific gap-junction coupling and yields a comprehensive connexin-interaction map. We experimentally confirmed Connexin 43-mediated interfaces among astrocytes, microglia and endothelial cells, which were among the highest- ranked predictions; in contrast, existing approaches fail to prioritize these confirmed interactions. Differential analysis of an Alzheimers disease model versus wild-type identifies preserved astroglial coupling but increased microglia-associated gap-junction communication. In summary, CellWHISPER disentangles direct cell-cell communication from structural proximity, enabling statistically robust and computationally scalable inference from spatial transcriptomics, leading to the discovery of tissue- and disease-associated contact-mediated signaling programs.

bioinformatics↗
bioRxiv2026

Cell-Permeable Morpholino Inhibits Programmed Death-Ligand 1 at mRNA Level and Potentiates Antitumor Immunity in Breast Cancer

Morpholino-based ASOs with nuclease-resistant backbone and high sequence specificity could modulate the PD-L1/PD-1 axis to overcome T regulatory cell mediated inhibition but poor cell-permeability limits their clinical applications. Here, we develop a guanidinium-modified cell-permeable morpholino specifically targeting PD-L1 in breast cancer. It suppresses PD-L1 expression through effectively mitigating cytokine induced PD-L1 upregulation and enhancing autophagic flux. Notably, the ASO synergistically with durvalumab (anti-PD-L1 antibody) disrupt, hyperglycosylated PD-L1 by mRNA-level interference and also reduced PD-1 T-cell populations. Importantly, systemic and intra-tumoral administration of ASO demonstrated no abnormalities in cardiac tissues which show no off-target toxicity but it reprogrammed the tumor microenvironment by upregulating inducible nitric-oxide synthase and suppressed regulatory cells, collectively enhancing antitumor immune responses. Together, these findings establish morpholino ASO as a promising targeting therapeutic strategy to overcome tumor-induced immunosuppression, opening new avenues for precision treatment in breast-cancer.

cancer biology↗
bioRxiv2025

COMPASS: A Web-Based COMPosite Activity Scoring System to Navigate Health and Disease Through Deterministic Digital Biomarkers

Reproducible and clinically transferable quantification of pathway activity remains a major barrier in precision medicine, where biomarker performance often depends on cohort composition and normalization strategies. Here, we introduce COMPASS (COMPosite Activity Scoring System), a deterministic threshold-based framework that converts gene expression into quantitative pathway activity scores without reliance on reference cohorts. COMPASS derives gene-specific activation thresholds directly from data, standardizes deviations from thresholds, and integrates directionally opposing genes into a single composite score. This enabled transparent activity scoring, statistical comparisons, and survival analyses without coding. Across diverse biological and clinical datasets, COMPASS robustly quantified cellular states, benchmarked the humanness and disease relevance of new approach methodologies, and stratified outcomes. Compared to GSVA and ssGSEA, COMPASS demonstrated greater consistency across datasets and improved robustness in bootstrap analyses, particularly for bidirectional programs, including regulatory-approved sepsis gene signatures. COMPASS therefore addresses a critical unmet need for exact, interpretable, and clinically transferable biomarker discovery and outcome modeling across diverse biological and clinical settings.

bioinformatics↗
bioRxiv2025

Optimal bioelectric control accelerates collective wound healing

Collective cell migration is critical in a range of biophysical processes spanning wound healing to tumor metastasis. It is therefore important to develop techniques for regulating cell motion, but while we have developed powerful migration tools from optogenetics and bioelectricity, the complex mechanics of collective systems make it difficult to determine where and when to apply these stimuli. For example, here we begin with a circular sheet of skin cells with a central hole, or "wound", and show that globally stimulating all cells to migrate radially inwards to close the gap through electrotaxis causes catastrophic mechanical damage, emerging from strong cell-cell adhesion, which we explain with an active elasticity model. We propose a solution inspired by sheepherding based on using local stimulation to learn the collective perturbation-response of the group and then drive tissue motion. First, we induce local electrotaxis to characterize the impulse-response function of quasi-1D strips of skin tissue, discovering that hyper-local stimulation triggers long-range tissue response over a length scale set by cell-cell adhesion also predicted by our model. Based on this, we apply local, concentric ring fields to in vitro circular wounds. Continuously driving cells towards the wound core kinetically traps the tissue in a jammed state and freezes healing. Pulsing the stimuli allows the tissue to relax and fluidize again, accelerating migration. Finally, we integrate the key length and timescales of tissue mechanics into a biophysically-informed continuum control model. The models predictive framework helps determine where and when to apply stimulation for optimal tissue growth which, when tested, accelerates healing 5-fold.

biophysics↗
bioRxiv2025

Chemical Dice Integrator (CDI): A Scalable Framework for Multimodal Molecular Representation Learning

The machine learning landscape for molecular property prediction is fragmented, with numerous Featurizers each capturing a narrow, specialized view of chemical structure. This heterogeneity forces a suboptimal choice of representation a priori, limiting model generalizability. We introduce the Chemical Dice Integrator (CDI), a hierarchical framework that unifies six orthogonal molecular representations, physicochemical (Mordred), topological (GROVER), visual (ImageMol), biological (Signaturizer), quantum-mechanical (MOPAC), and linguistic (ChemBERTa), into a single, coherent embedding. The framework consists of CDI-Basic, a two-tiered autoencoder that fuses these modalities, and CDI-Generalised, a Mamba State-Space Model (SSM) that learns a direct, efficient map from SMILES strings to the unified embedding space. Extensive benchmarking across 23 classification (171 tasks) and 10 regression datasets demonstrates that CDI embeddings consistently achieve superior predictive performance compared to individual Featurizers and standard feature aggregation methods. The CDI-Generalised model achieves this performance with exceptional computational efficiency, outperforming deep learning Featurizers in terms of speed and resource overhead. Furthermore, we demonstrate that the CDI embedding is chemically intuitive, allowing for the sensitive distinction of nuanced structural variants, such as chiral enantiomers and kekulized SMILES forms. By bridging multimodal chemical intelligence with scalable, sequence-based inference, CDI offers a strong foundation for molecular machine learning.

bioinformatics↗
bioRxiv2025

Atypical activation and molecular glue-like dimerization mechanism of an intrinsically-biased chemokine receptor

CXCR7, also known as atypical chemokine receptor 3 (ACKR3), is a naturally-biased, {beta}-arrestin-coupled seven transmembrane receptor (7TMR) that lacks productive coupling with heterotrimeric G-proteins. Despite a critical involvement in cancer metastasis, cardiovascular pathophysiology, and inflammatory disorders, the molecular basis of non-canonical activation and functional divergence of CXCR7 remains elusive. Here, we present a complete landscape of CXCR7 activation using a series of cryo-EM structures, and discover an atypical activation mechanism that is distinct from prototypical GPCRs. CXCR7 is maintained in a basal conformation by a unique tripartite ionic-lock involving TM5-TM6, in contrast to a broadly conserved TM3-TM6 ionic-lock in GPCRs, which is disrupted upon receptor activation. Importantly, activation of CXCR7 results in a constricted pocket and distinct surface topology on the intracellular side compared to prototypical GPCRs. Serendipitously, we capture novel dimeric arrangements of CXCR7 with an inter-protomer stitching by a native phospholipid serving as a molecular glue, and identify previously unanticipated extrahelical allosteric sites on the receptor. Surprisingly, in an intermediate state structure of CXCR7, the second extracellular loop (ECL2) displays a self-blocking conformation, in stark contrast to ECL2-mediated self-activating mechanism reported recently for some orphan GPCRs. Finally, we unequivocally establish CXCR7 as an atypical opioid receptor via a large peptide library screening and structure elucidation in complex with distinct opioid peptides imparting full receptor activation. In summary, our study elucidates an atypical mechanism of CXCR7 activation, and establishes it as an alternative, non-canonical opioid receptor target with potential for novel pain therapeutics.

biochemistry↗
bioRxiv2025

A simple circuit to sustain intact tumor microenvironments for complex drug interrogations

Deep learning and large language models can integrate complex datasets to uncover biological insights that are often undetectable through conventional analyses. With application to translational cancer research, these computational tools have positioned 3D patient-derived tumor avatars front and center as crucial data input sources. However, a major challenge remains: the lack of standardization in media composition in 3D patient-derived tumor models unpredictably affects cell behavior and limit the utility beyond predicting treatment responses. To address this unmet need, we developed a simple, reproducible perfusion circuit system to approximate in vivo physiology using autologous patient plasma. With peritoneal metastases and core needle biopsies across multiple tumor histologies, we demonstrate preservation of the tumor microenvironment for up to 48 hours using multi-modal interrogation techniques. With proof-of-concept experiments, we display the systems ability to unveil complex drug-dependent biology within this time window. Standardizable, physiologically relevant platforms for 3D patient-derived tumor avatars will yield unprecedented insights through the integration of data from broad groups of patients and the use of an expanding armamentarium of artificial intelligence capabilities.

cancer biology↗
bioRxiv2025

Mapping Structural Aging across Human Tissues reveals tissue-specific trajectories, coordinated deterioration and genetic determinants

1.Tissue structure, the organization of cells, vasculature and extracellular matrix, underpins organ function. Yet how it deteriorates with age remains largely uncharacterized. Current aging research focuses primarily on molecular changes, missing this structural dimension. Here we present PathStAR (Pathology-based Structural Aging Rate), a computational framework that quantifies tissue structural aging from histopathology images without training on chronological age. Applying PathStAR to 25,306 postmortem biopsies spanning 40 tissue types from 970 individuals aged 21-70, we show that structural aging unfolds through non-linear phases rather than gradual decline. We identify three distinct temporal programs: early-aging tissues (vascular system, peaking in the 30s), late-aging tissues (uterus and vagina, peaking around menopause) and biphasic-aging tissues (digestive and male reproductive organs, with two acceleration periods). During accelerated phases, tissues exhibit common molecular signature of increased inflammation, decreased energy production, regeneration and quality control alongside tissue-specific pathway disruptions related to its function: Artery-specific decline of peroxisomal function, responsible for fatty acid breakdown and testis-specific decline of spermatogenesis. Cross-organ analysis reveals coordinated deterioration within individuals, not only within expected organ systems but also unexpectedly between digestive and reproductive tissues, traced to shared sex hormone signaling. This reveals a role for sex hormones in maintaining structural integrity of non-reproductive organs during aging. Genome-wide association analysis identifies 123 germline variants associated with organ-specific accelerated structural aging, including variants in the longevity regulator SIRT6 linked specifically to vascular structural decline. As proof of concept, PathStAR captures the established non-linear functional decline of the ovary, fertility loss in the 30s and menopause in the 50s, that bulk transcriptomic and methylation profiles from matched samples fail to detect. PathStAR provides a systematic map of structural aging across the human body: when tissues structure decline, what molecular programs define each phase, and how organs deteriorate in coordination.

bioinformatics↗
bioRxiv2025

PathQC: Determining Molecular and Physical Integrity of Tissues from Histopathological Slides

Quantifying tissue molecular and physical integrity is essential for biobank development. However, current assessment methods either involve destructive testing that depletes valuable biospecimens or rely on manual evaluations, which are not scalable and lead to interindividual variation. To overcome these challenges, we present PathQC, a deep learning framework that directly predicts the tissue RNA Integrity Number (RIN) and the extent of autolysis from hematoxylin and eosin (H&E)-stained whole-slide images of normal tissue biopsies. PathQC first extracts morphological features from the slide using a recently developed digital pathology foundation model (UNI), followed by a supervised model that learns to predict RNA Integrity Number and autolysis scores from these morphological features. PathQC is trained on and applied to the Genotype-Tissue Expression (GTEx) cohort, which comprises 25,306 non-diseased post-mortem samples across 29 tissues from 970 donors, where paired ground truth RIN and autolysis scores were available. Here, PathQC predicted RIN with an average correlation of 0.47 and an autolysis score of 0.45, with notably high performance in Adrenal Gland tissue (R=0.82) for RIN and in Colon tissue (R=0.83) for autolysis. We provide a pan-tissue model for the prediction of RIN and autolysis score for a new slide from any tissue type (GITHUB). Overall, PathQC will enable scalable measurement of molecular and physical integrity from routine H&E images, thereby enhancing the quality of both biobank generation and its retrospective analysis.

bioinformatics↗
bioRxiv2025

Multi-omics integration at cell type resolution uncovers gene-metabolite mechanisms underlying osteoarthritis heterogeneity

Metabolic dysregulation is an important factor for osteoarthritis pathogenesis, but comprehensive studies of underlying mechanisms and pathways are rare. We analyzed newly generated metabolomics data on bone marrow from 119 osteoarthritis patients, along with single-cell transcriptomics data to reconstruct networks of gene-metabolite associations at cell type resolution. Hubs of these networks - cell type-specific as well as pan-cell type hubs - revealed key molecular factors of osteoarthritis heterogeneity. Systems-level analysis of hubs revealed major roles for glycerophospholipid, glycerolipid and sphingolipid metabolism pathways, along with lipid signaling. We used Machine Learning models of gene-metabolite relationships to discover cell types most relevant to each metabolite. Integrative analysis of disease severity scores along with multi-omics data revealed a shift in specific immune cell subtypes in low versus high grade disease. We conclude that leveraging gene-metabolite covariation in a patient cohort can uncover underlying molecular mechanisms, overcoming the challenges posed by high dimensionality of multi-omics data.

bioinformatics↗