Biology subjects

Goyal, Y.

Publications and source records attributed to Goyal, Y..

At least 19 recordsLinked to original sources

bioRxiv2026

ExPert infers causal perturbations from cell states across biological and clinical contexts

Pooled single-cell CRISPR screens have driven substantial effort toward models predicting the response to a perturbation. We posed the inverse question: which perturbation produced an observed cell state? Here we developed ExPert (Extracting Perturbations), which learns one prototype per perturbation in a latent space shared by 12 cellular contexts, trained on 8,964 perturbations and 19.5 million cells, and ranks 19,895 candidate genes per query, over half absent from training. ExPert more than doubled held-out-context recall over the best existing method with 1,000-fold less compute, and transferred without retraining to drugs, protein degradation and combinations. In alcohol-associated liver disease it implicated mRNA export, unchanged in expression, and we experimentally confirmed nuclear poly(A) retention and isoform switching. In primary mitochondrial disease it recovered the affected complex from blood, localized the signal to monocytes, and flagged a misdiagnosed control. Perturbation atlases can thus be queried in reverse, from state to likely driver.

systems biology↗

bioRxiv2026

Palmitate-activated IRE1-XBP1 signaling limits DNA damage and is associated with H2A.X-linked DNA damage response transcripts in triple-negative breast cancer cells

Palmitate-associated lipotoxic stress activates the endoplasmic-reticulum stress sensor protein IRE1, but how IRE1 signaling influences genotoxic-stress responses in triple-negative breast cancer remains unclear. Here, we investigated IRE1 and XBP1 signaling in MDA-MB-231 cells exposed to palmitate and etoposide. Under combined lipotoxic and genotoxic stress, XBP1 depletion increased DNA-damage accumulation in IRE1 wild-type cells but not in IRE1-deficient cells, whereas IRE1 re-expression reduced DNA damage. Pharmacologic IRE1 RNase inhibition also increased etoposide-associated DNA damage, supporting a protective role for IRE1 RNase-dependent signaling. IRE1 status and XBP1 manipulation differentially affected H2A.X and {gamma}H2A.X abundance, indicating context-dependent regulation of chromatin-associated DNA-damage signaling. At the RNA level, H2A.X was enriched in IRE1 immunoprecipitates, and the H2A.X-linked DNA-damage-response transcripts HUWE1 and BRCA2 also associated with IRE1 under combined stress. These findings link palmitate-activated IRE1-XBP1 signaling to genome-protective responses during genotoxic stress and identify a network of IRE1-associated DNA-damage-response transcripts in triple-negative breast cancer cells. Direct IRE1-mediated cleavage of these transcripts remains to be established.

cancer biology↗

bioRxiv2026

Leveraging Twin Information for Inference of Gene Regulatory Networks

Gene regulatory networks (GRNs) underlie maintenance of cellular phenotypes and responses to stimuli. Modern single-cell profiling methods offer high-throughput datasets to infer GRNs en masse but do not capture dynamic information. Cell-state heterogeneity further confounds correlation-based inference approaches. We addressed these challenges with TwINFER, a conceptual framework leveraging information from recently divided sister cells, or "twins," identifiable via recently developed barcoding techniques. We show that twin information discriminates regulatory from non-regulatory correlations and resolves interaction direction and type (activation/repression). We performed a diverse set of simulations, covering common network motifs and large-scale networks, where TwINFER outperforms state-of-the-art inference capabilities. Crucially, TwINFER resolved the commonly observed false positives in fan-out and feed-forward loop motifs where most methods perform poorly. Lastly, we applied TwINFER to a lineage-barcoded hematopoiesis dataset which refined the network inference, flagged multi-state genes, and determined causal relations. Our work exploits cellular twins as untapped information, readily complementing existing inference approaches.

systems biology↗

bioRxiv2025

Interpretable Thermodynamic Score-based Classification of Relaxation Excursions

Classifying cellular states from high-dimensional molecular and genomic measurements requires methods that provide not only accurate predictions but also calibrated uncertainty and interpretability. Current nonlinear classifiers offer accuracy but often lack uncertainty quantification and mechanistic insights into the features that matter most. We introduce Keeping SCORE, a framework that transforms conditional diffusion models into probabilistic engines for classification and regression by computing exact likelihoods along stochastic noising trajectories. We first benchmark Keeping SCORE on image recognition tasks (handwritten digits, natural photos). We then apply Keeping SCORE to single-cell transcriptomics across a 22-million-cell atlas, classifying 164 cell types with accuracy matching or exceeding state-of-the-art methods, while uniquely providing posterior probability estimates and prediction confidence. For genetic perturbation mapping across 100 CRISPRi conditions in a multi-study Perturb-seq dataset, our approach again matches or surpasses discriminative baselines, with feature-level attributions identifying which genomic features drive each decision. Applied to large-scale protein sequence data, our framework accurately regresses mutational stability effects, attributing them quantitatively to positions along the input sequence. Keeping SCORE requires no retraining or architectural changes to existing diffusion models, providing portable, interpretable, and uncertainty-aware predictions for biological discovery.

bioinformatics↗

bioRxiv2025

Fluctuation structure predicts genome-wide perturbation outcomes

Pooled single-cell perturbation screens represent powerful experimental platforms for functional genomics, yet interpreting these rich datasets for meaningful biological conclusions remains challenging. Most current methods fall at one of two extremes: either opaque deep learning models that obscure biological meaning, or simplified frameworks that treat genes as isolated units. As such, these approaches overlook a crucial insight: gene co-fluctuations in unperturbed cellular states can be harnessed to model perturbation responses. Here we present CIPHER (Covariance Inference for Perturbation and High-dimensional Expression Response), a conceptual framework leveraging ideas from linear response in statistical physics to model transcriptome-wide perturbation outcomes using gene co-fluctuations in unperturbed cells. We validated our approach on synthetic regulatory networks before applying it to 29 large-scale single-cell genetic perturbation datasets covering 19,003 perturbations and over 6.26M cells. Our work robustly recapitulated genome-wide responses to single and double perturbations by exploiting baseline gene covariance structure. Importantly, eliminating gene-gene covariances, while retaining gene-intrinsic variances, i.e. mean-field conditions, dramatically reduced model performance by several folds across multiple metrics, demonstrating the rich information stored within baseline fluctuation structures. Benchmarked against recent deep learning and linear baselines, CIPHER matched or exceeded the best-performing approaches with fitting a single parameter. Moreover, gene-gene correlations transferred successfully across independent studies of the same cell type, revealing stereotypic fluctuation structures. We further extended CIPHER to the inverse problem of identifying true driver perturbations, where it achieved high performance across both genetic and chemical perturbation screens through uncertainty-aware Bayesian inference. We further used CIPHER to nominate drivers of therapy resistance in melanoma and pancreatic cancer, validating its top predictions experimentally. Finally, most genome-wide responses propagated through the covariance matrix along approximately 1-3 independent and global gene modules, consistent with a low-rank structure of the underlying gene regulatory network, which we show can enable the framework's success. We have also created a package called cipher-perturb, available on PyPI, to apply the framework to any dataset, accompanied by a detailed website (https://goyallab.github.io/CIPHERWebsite/). Our study underscores the importance of theoretically-grounded models in capturing complex biological responses, highlighting fundamental design principles encoded in cellular fluctuation patterns.

systems biology↗

bioRxiv2025

Failed Metabolic Adaptation to Stress Underlies Pathogenesis in a Heterozygous Genetic Disorder

Disorders linked to heterozygous variants occupy a continuum in terms of the timing and severity of phenotypic emergence. An important question regarding this variability entails the effect stress has on the residual protein function. Using Darier disease (DD), caused by heterozygous variants of the SERCA2 calcium pump, as a model, we uncovered a potential connection between extrinsic stress and pathogenesis. The skin lesions characteristic of DD entail loss of intercellular adhesion and rarely appear pre-adolescence, suggesting that factors beyond heterozygosity contribute to disease pathogenesis. Testing whether age-related stressors contribute to DD, we show that DD patient-derived keratinocytes subjected to stress yield twice the reactive oxygen species of controls, accompanied by greater disruption of intercellular adhesion. Metabolic analysis of DD cells revealed perturbation of the pentose phosphate pathway (PPP), a stress response system responsible for regenerating antioxidants like glutathione. At baseline, DD cells had less free glutathione but an increase in protective glutathione-based modifications of SERCA2, a reversible form of protein oxidation. With stress, DD cells form an aberrant, heavily glutathionylated perinuclear halo consisting of keratin and the intercellular adhesion component, desmoplakin. We propose a model whereby SERCA2 heterozygosity causes mild oxidative stress that under homeostatic conditions can be buffered by glutathionylation. When stressed, the depleted glutathione store is shunted towards the desmoplakin-intermediate filament system at the expense of SERCA2, rendering it vulnerable to damage. A lesional flare, then, would represent a case of more complete SERCA2 inhibition and a novel example of how heterozygous disorders interact with stress to disrupt intercellular adhesion.

cell biology↗

bioRxiv2024

Resolving the design principles that control postnatal vascular growth and scaling

After birth, tissues grow continuously until reaching adult size, with each organ exhibiting unique cellular dynamics, growth patterns, and (stem or non-stem) cell sources. Using a suite of experimental and computational multiscale approaches, we found that aortic expansion is guided by specific biological principles and scales with the vertebral column rather than animal body weight. Expansion proceeds via two distinct waves of arterial cell proliferation along blood flow that are spatially stochastic, yet temporally coordinated. Each wave exhibits unique cell cycle kinetics and properties, with the first wave exhibiting cell cycle durations as fast as 6 hours. Single-cell RNA sequencing showed changes in fatty acid metabolism concomitant with an increase in cell size. Mathematical modeling and experiments indicated endothelial cell extrusion is essential for homeostatic aortic growth and balancing excess proliferation. In a genetic model of achondroplasia, the aorta achieves proper scaling through enhanced cell extrusion while maintaining normal proliferation dynamics. Collectively, these results provide a blueprint of the principles that orchestrate aortic growth which depends entirely on differentiated cell proliferation rather than resident stem cells.

developmental biology↗

bioRxiv2024

Dysregulated RNA splicing induces regeneration failure in alcohol-associated liver disease

Individuals with progressive liver failure are at a high risk of mortality without liver transplantation. However, our understanding of derailed regenerative responses in failing livers is limited. Here, we performed comprehensive multi-omic profiling of healthy and diseased human livers using bulk and single-nucleus RNA-plus ATAC-seq. We report that hepatic immune milieu alterations in alcohol-associated liver disease (ALD) prevent hepatocytes from transitioning to a proliferative progenitor-like state, trapping them into an unproductive intermediate state. We discovered striking changes in RNA binding protein (RBP) expression, particularly ESRP, PTBP, and SR families, that cause misregulation of developmentally controlled RNA splicing in ALD. Our data pinpoint ESRP2 as a pivotal disease-sensitive RBP and support a causal role of its deficiency in ALD pathogenesis. Notably, splicing defects in ESRP2-targets Tcf4 and Slk, amongst others, directly alter their nuclear localization and activities, disrupting WNT and Hippo signaling pathways, which are critical for normal liver regeneration. We demonstrate that changes in stromal cell populations enrich failing ALD livers with TGF-{beta}, which suppresses ESRP2-driven epithelial splicing program and replaces functional parenchyma with quasi-progenitor-like cells lacking liver-specific functions. This unprecedented account of transcriptional and post-transcriptional dysregulation in ALD suggests that targeting misspliced RNAs could improve recovery and serve as biomarkers for poor ALD outcomes.

molecular biology↗

bioRxiv2024

AP-1 Mediates Cellular Adaptation and Memory Formation During Therapy Resistance

Cellular responses to environmental stimuli are typically thought to be governed by genetically encoded programs. We demonstrate that melanoma cells can form and maintain cellular memories during the acquisition of therapy resistance that exhibit characteristics of cellular learning and are dependent on the transcription factor AP-1. We show that cells exposed to a low dose of therapy adapt to become resistant to a high dose, demonstrating that resistance was not purely selective. The application of therapy itself results in the encoding of transient gene expression into cellular memory and that this encoding occurs for both transiently induced and probabilistically arising expression. Chromatin accessibility showed concomitant persistence. A two-color AP-1 reporter system showed that these memories are encoded in cis, constituting an example of activating cis epigenetics. Our findings establish the formation and maintenance of cellular memories as a critical aspect of gene regulation during the development of therapy resistance.

systems biology↗

bioRxiv2024

Piscis: a novel loss estimator of the F1 score enables accurate spot detection in fluorescence microscopy images via deep learning

Single-molecule RNA fluorescence in situ hybridization (RNA FISH)-based spatial transcriptomics methods have enabled the accurate quantification of gene expression at single-cell resolution by visualizing transcripts as diffraction-limited spots. While these methods generally scale to large samples, image analysis remains challenging, often requiring manual parameter tuning. We present Piscis, a fully automatic deep learning algorithm for spot detection trained using a novel loss function, the SmoothF1 loss, that approximates the F1 score to directly penalize false positives and false negatives but remains differentiable and hence usable for training by deep learning approaches. Piscis was trained and tested on a diverse dataset composed of 358 manually annotated experimental RNA FISH images representing multiple cell types and 240 additional synthetic images. Piscis outperforms other state-of-the-art spot detection methods, enabling accurate, high-throughput analysis of RNA FISH-derived imaging data without the need for manual parameter tuning.

bioinformatics↗

bioRxiv2023

TGF-β blockade drives a transitional effector phenotype in T cells reversing SIV latency and decreasing SIV reservoirs in vivo

HIV-1 persistence during ART is due to the establishment of long-lived viral reservoirs in resting immune cells. Using an NHP model of barcoded SIVmac239 intravenous infection and therapeutic dosing of the anti-TGFBR1 inhibitor galunisertib (LY2157299), we confirmed the latency reversal properties of in vivo TGF-{beta} blockade, decreased viral reservoirs and stimulated immune responses. Eight SIV-infected macaques on ART were treated with four 2-week cycles of galunisertib. ART was discontinued 3 weeks after the last dose, and macaques euthanized 6 weeks after ART-interruption (ATI). 7 out of 8 macaques rebounded between week 2 and 6 post-ATI. Galunisertib led to viral reactivation as indicated by plasma viral load and immunoPET/CT with 64Cu-DOTA-F(ab)2-p7D3-probe. A decrease in cell-associated (CA-)SIV DNA was detected in lymph nodes, gut and PBMC, while intact pro-virus in PBMC decreased by 3-fold. No systemic increase in inflammatory cytokines was observed. High-dimensions cytometry, bulk, and single-cell (sc)RNAseq revealed a shift toward an effector phenotype in T and NK cells characterized by a progressive downregulation in TCF1. In summary, we demonstrated that galunisertib, a clinical stage TGF-{beta} inhibitor, reverses SIV latency and decreases SIV reservoirs by driving T cells toward an effector phenotype, enhancing immune responses in vivo in absence of toxicity. One-sentence summaryTGF-{beta} blockade drives an effector phenotype in immune cells leading to SIV latency reversal and enhanced immune responses in vivo.

immunology↗

bioRxiv2023

Prevalence of and gene regulatory constraints on transcriptional adaptation in single cells

Cells and tissues have a remarkable ability to adapt to genetic perturbations via a variety of molecular mechanisms. Nonsense-induced transcriptional compensation, a form of transcriptional adaptation, has recently emerged as one such mechanism, in which nonsense mutations in a gene can trigger upregulation of related genes, possibly conferring robustness at cellular and organismal levels. However, beyond a handful of developmental contexts and curated sets of genes, to date, no comprehensive genome-wide investigation of this behavior has been undertaken for mammalian cell types and contexts. Moreover, how the regulatory-level effects of inherently stochastic compensatory gene networks contribute to phenotypic penetrance in single cells remains unclear. Here we combine computational analysis of existing datasets with stochastic mathematical modeling and machine learning to uncover the widespread prevalence of transcriptional adaptation in mammalian systems and the diverse single-cell manifestations of minimal compensatory gene networks. Regulon gene expression analysis of a pooled single-cell genetic perturbation dataset recapitulates important model predictions. Our integrative approach uncovers several putative hits--genes demonstrating possible transcriptional adaptation--to follow up on experimentally, and provides a formal quantitative framework to test and refine models of transcriptional adaptation.

systems biology↗

bioRxiv2023

singletCode: synthetic barcodes identifies singlets and evaluates doublet algorithms in scRNA-seq datasets

Single-cell RNA sequencing datasets comprise true single cells, or singlets, in addition to cells that coalesce during the protocol, or doublets. Identifying singlets with high fidelity in single-cell RNA sequencing is necessary to avoid false negative and false positive discoveries. Although several methodologies have been proposed to infer true singlets and doublets, they typically rely on datasets being highly heterogeneous. Here we develop and apply singletCode, a computational framework that leverages datasets with synthetically introduced DNA barcodes for a hitherto unexplored application: to extract ground truth singlets. We demonstrate the feasibility of singlets extracted via singletCode to evaluate the performance and robustness of existing doublet detection methodologies. We find that existing doublet detection methods show lower than documented sensitivity to doublets simulated with ground truth singlets. As DNA barcoded datasets are being increasingly reported, singletCode can identify singlets and inform rational choice of doublet detecting algorithms and their associated limitations.

systems biology↗

bioRxiv2023

Retrospective identification of intrinsic factors that mark pluripotency potential in rare somatic cells

Pluripotency can be induced in somatic cells by the expression of the four "Yamanaka" factors OCT4, KLF4, SOX2, and MYC. However, even in homogeneous conditions, usually only a rare subset of cells admit reprogramming, and the molecular characteristics of this subset remain unknown. Here, we apply retrospective clone tracing to identify and characterize the individual human fibroblast cells that are primed for reprogramming. These fibroblasts showed markers of increased cell cycle speed and decreased fibroblast activation. Knockdown of a fibroblast activation factor identified by our analysis led to increased reprogramming efficiency, identifying it as a barrier to reprogramming. Changing the frequency of reprogramming by inhibiting the activity of LSD1 led to an enlarging of the pool of cells that were primed for reprogramming. Our results show that even homogeneous cell populations can exhibit heritable molecular variability that can dictate whether individual rare cells will reprogram or not.

systems biology↗

bioRxiv2022

Metastatic potential in clonal melanoma cells is driven by a rare, early-invading subpopulation

Metastasis occurs when tumor cells leave the primary tumor site and disseminate to distal organs. Even though most cells remain in the primary tumor, the circumstances by which a small fraction of them disseminate remain unclear. Here, we show that a rare, highly invasive subpopulation of melanoma cells can be detected within clonal cell lines due to non-genetic fluctuations in gene expression. The highly invasive phenotype was intrinsic to the cells, independent of their environment, and was marked by transiently high levels of SEMA3C expression, as revealed by RNA-sequencing analysis. Furthermore, the invasive subpopulation drove the bulk dissemination of tumor cells to distal locations in a mouse model of melanoma. The transcription factor NKX2.2 regulated the proportion of invasive cells in the melanoma 1205Lu cell line. Furthermore, an overall tradeoff between proliferation and invasion in single cells was observed. Our results suggest that phenotypes like metastasis may arise from intrinsic differences stemming from non-genetic fluctuations between single cells.

cancer biology↗

bioRxiv2022

clampFISH 2.0 enables rapid, scalable amplified RNA detection in situ

RNA labeling in situ has enormous potential to visualize transcripts and quantify their levels in single cells, but it remains challenging to produce high levels of signal while also enabling multiplexed detection of multiple RNA species simultaneously. Here, we describe clampFISH 2.0, a method that uses an inverted padlock design to efficiently detect and exponentially amplify signals from many RNA species at once, while also reducing time and cost compared to the prior clampFISH method. We leverage the increased throughput afforded by multiplexed signal amplification and sequential detection by demonstrating the ability to detect 10 different RNA species in over 1 million cells. We also show that clampFISH 2.0 works in tissue sections. We expect the advantages offered by clampFISH 2.0 will enable many applications in spatial transcriptomics.

bioengineering↗

bioRxiv2022

ClonoCluster: a method for using clonal origin to inform transcriptome clustering

Clustering cells based on their high dimensional profiles is an important data reduction process by which researchers infer distinct categories of cellular state. The advent of cellular barcoding, however, provides an alternative means by which to group cells: by their clonal origin. We developed ClonoCluster, a computational method that combines both clone and transcriptome information to create hybrid clusters that weight both kinds of data with a tunable parameter. We generated hybrid clusters across six independent datasets and found that ClonoCluster generated qualitatively different clusters in all cases. The markers of these hybrid clusters were different but had equivalent fidelity to transcriptome-only clusters. The genes most strongly associated with the rearrangements in hybrid clusters were ribosomal function and extracellular matrix genes. We also developed the complementary tool Warp Factor that incorporates clone information in popular 2D visualization techniques like UMAP. Integrating ClonoCluster and Warp Factor revealed biologically relevant markers of cell identity.

bioinformatics↗

bioRxiv2022

Changes in chromatin accessibility are not concordant with transcriptional changes for single-factor perturbations

A major goal in the field of transcriptional regulation is the mapping of changes in the binding of transcription factors to the resultant changes in gene expression. Recently, methods for measuring chromatin accessibility have enabled us to measure changes in accessibility across the genome, which are thought to correspond to transcription factor binding events. In concert with RNA-sequencing, these data in principle enable such mappings; however, few studies have looked at their concordance over short duration treatments with specific perturbations. Here, we used tandem, bulk ATAC-seq and RNA-seq measurements from MCF-7 breast carcinoma cells to systematically evaluate the concordance between changes in accessibility and changes in expression in response to retinoic acid and TGF-{beta}. We found two classes of genes whose expression showed a significant change: those that showed some change in accessibility of nearby chromatin, and those that showed virtually no change despite strong changes in expression. The peaks associated with genes in the former group had a lower baseline accessibility prior to exposure to signal. Focusing the analysis specifically on peaks with motifs for transcription factors associated with retinoic acid and TGF-{beta} signaling did not reduce the lack of correspondence. Analysis of paired chromatin accessibility and gene expression data from distinct paths along the hematopoietic differentiation trajectory showed a much stronger correspondence, suggesting that the multifactorial biological processes associated with differentiation may lead to changes in chromatin accessibility that reflect rather than drive altered transcriptional status. Together, these results show many gene expression changes can happen independent of changes in accessibility of local chromatin in the context of a single-factor perturbation and suggest that some changes to accessibility changes may occur after changes to expression, rather than before.

bioinformatics↗